Understanding the impact of AI in the advice sector

0:09
Hi, welcome, everyone.

0:11
We'll be starting at 1:30.

0:12
We're just going to let everyone sort of start flowing into the room.

1:06
Welcome to everyone that's just joined.

1:08
Got lots of people coming in now.

1:10
We're going to be starting at 1:30, just in a couple of minutes time.

2:26
For those that have just joined, welcome.

2:28
We've got over 200 people in the room now.

2:30
We're going to be starting in just a couple of minutes.

3:29
Great.

3:29
Welcome, everyone.

3:31
We're still going to just leave it a couple minutes more just to let everyone kind of get into the room.

4:19
Great.

4:19
Welcome, everyone.

4:20
I know it's 1:30, but we'll just give it another few seconds.

4:23
I can see there's lots of people still joining.

5:00
OK, great.

5:01
I can steal this.

5:01
A few more people coming in.

5:04
Just give it another couple more seconds.

5:07
We've had so many of you register.

5:08
It's been such an exciting event.

5:29
OK, great.

5:30
Let's kick off.

5:31
I know there's a few more people still coming in, but we're at the 1:30 mark now.

5:35
So first of all, massive welcome to you all.

5:38
My name is France Little, I'm the AI coordinator at Child Poverty Action Group, and I'm delighted to be hosting this webinar where we're going to be looking at understanding the impact of AI in the advice sector.

5:49
How this webinar is going to work is we're going to have about sort of an hour, just under an hour of a panel discussion.

5:55
And I'll introduce my panellists in a moment, but then we'll also leave lots of time for questions.

6:00
We want to leave a good 20-30 minutes.

6:02
So really encourage you to sort of add your questions to the Q&A and continue to add them as we discuss.

6:09
And we'll try and get to as many as we can.

6:14
But before we get going, just a bit of housekeeping.

6:17
First one to say is that this is being recorded and we are going to be putting this recording onto our YouTube channel.

6:23
So again, if you've got colleagues who haven't been able to make it today, please do that, direct them to our YouTube.

6:31
We'll also, if you have any technical issues today, just let us know.

6:36
The best way to do that is to e-mail the training team at Training at cpag.org.uk and someone is monitoring that inbox and we'll be able to assist you immediately.

6:47
If you'd like to ask a question, please use the Q&A function.

6:50
You'll find that in the toolbar.

6:52
And again, we'll try and get to as many questions as we can.

6:56
I would encourage you, there's lots of people on the call.

6:59
Do have a read through of what other people have asked before adding your own just to make sure you're not duplicating.

7:04
It just really helps us in helping to manage it.

7:07
And yeah, as you'll see as well, the chat function is disabled, so please only use the Q&A.

7:14
The other aspect of that as well, of course, is you can thumbs up.

7:17
So if you go into the toolbar, you'll see you can click a little thumbs and up vote.

7:21
And again, that just helps us to understand which is the most pressing questions.

7:25
So with that housekeeping out the way, I actually think it's probably worthwhile just doing a quick definition of what AI is, because let's not assume that everyone knows.

7:37
So AI, of course, stands for artificial intelligence.

7:40
So in brief, it's a term that describes a set of technologies that carry out tasks commonly thought to require intelligence.

7:48
That can include things like recognising patterns, processing language, problem solving, or even decision making.

7:54
And AI systems typically analyse large amounts of data to be able to carry out these kinds of tasks.

8:01
So when we're talking about AI, there's kind of three key things to kind of pick out from this definition.

8:06
The first is that AI is a term that's often used in a quite general sense, but it refers to a set of technologies, and you might be familiar with some of those technologies.

8:14
Things like genitive AI is the subset of AI that is used for things like chat bots like ChatGPT or Copilot.

8:21
But there's also things like natural language processing, which is the sort of AI that enables machines to learn language and communicate.

8:29
And that is in things like your voice operated GPSE systems like Google Maps or your voice assistants like Amazon's Alexa.

8:37
And then also you have things like computer vision.

8:39
So that's allowing machines to process, create and edit visual content like images and video.

8:45
And I'm sure you'll all have probably seen at some point now some form of AI generated image or video on social media and online generally.

8:53
So we're going to be using the term AI as the general term, but just know that we're going to be referring probably to all these different subsets.

9:01
The other key thing to point out here is that AI can complete tasks that are commonly thought to require intelligence.

9:08
So whilst with traditional computing we would normally have to kind of programme a machine in a way to specifically complete a task, what AI can do is it can learn patterns from training data and then use that knowledge to complete entirely new tasks that it's never seen before.

9:24
So think of it like when you put a question into something like ChatGPT or all of these chat bots, it can respond to your question even though it's never been directly programmed or even ever seen that question before.

9:36
And that kind of comes to the last kind of key bit to this definition.

9:39
And that's the fact that AI uses large amounts of data.

9:42
It needs large amounts of high quality data, preferably in order to be able to provide high quality responses to users.

9:49
And I'm sure we'll go into much more depth about all of that in a moment.

9:54
So why are we talking about this today?

9:55
Well, we had 900 + ups for this event, which I think definitely speaks for itself about why we need to talk about AI in the advice sector at the moment.

10:04
But we know that AI use is on the rise.

10:07
Recent reports show that it's doubled in the advice sector in the last year, and that in some ways comes as an unsurprising result.

10:14
We see there's huge pressures on the sector, there's a growing demand for expert advice.

10:19
Combined with things like falling budgets and economic pressures, it makes sense that technology like AI, which is promoted as this time saving tool, could actually support the sector in managing these ongoing challenges.

10:31
And we know that AI could offer lots of interesting, exciting opportunities from transcribing discussions between advisors and clients, generating case summaries to supporting advisors in their research and seeking out information.

10:46
So just even speeding up everyday tasks like scheduling or emails, but we also know there's a multitude of risks that we need to be thinking about maybe that security, that's how do we manage client data?

10:59
How do we ensure accuracy of the advice given?

11:02
And how do we manage the concerns that this about this replacing human judgement in high stakes situations as advice giving is.

11:11
And also thinking what does this mean about the level of trust and the relationship between the advice and client?

11:17
So This is why we're talking about this, and this is what we're going to be talking about in the next hour.

11:23
We'll be thinking about those opportunities and risks via what we're calling the client journey.

11:28
So from the moment a client decides that they need advice and they start searching online, they go beyond their family and friends to find an independent advice to them deciding to call up an agency and then receiving that advice.

11:41
We want to be thinking about where and how might AI be impacting this and what do you as an advisor, a supervisor and manager of an agency need to be thinking about in order to deliver high quality advice in the age of AI?

11:54
So with that, I'd love to introduce my panellists.

11:57
Today we have Dalava Warpitan Dal heads up the Advice U KS Membership and service department and he has over 18 years experience working in the social welfare, legal advice services and networks and has a particular focus on helping advice organisations engage with and respond to the impact of generative AI.

12:16
And he has a lot of expertise in organisational development, service design, policy assurance and risk management.

12:22
We also have Lindsay Paul here today, who's the Director of Advice Service Alliance, a role she's held for 12 years now.

12:30
And she began her career actually as a client and then as a volunteer for local Citizens Advice and has held many different roles within the advice sector, including chair of an early health justice partnership and a training development officer.

12:41
And in her mid career sought A venture into criminal justice, including several years as a government social researcher.

12:47
And she's then since returned to social welfare advice through short term roles and several pro bono charities.

12:53
And she's recently completed a sabbatical at the University of Oxford examining the relationship between pro bono and social welfare legal advice services.

13:00
So welcome, Welcome, Lindsay.

13:03
And last we have Baljeet Badesha, who has over 40 years of experience in the advice sector and has extensive experience running services, providing debt, housing and welfare advice services.

13:13
He's currently the director at the Nucleus Legal Advice and he's passionate about tackling poverty and inequality.

13:19
And Belgy has helped build partnerships between charities, local organisations and community groups to create inclusive, accessible and responsive services.

13:28
And he's also interested in exploring innovative approaches, including things like digital tools and AI to support and improve outcomes for clients.

13:36
He continues to champion social justice, empowerment and long term community well-being through collaboration and community engagement.

13:42
So welcome to you all and I'll just stop sharing so we can get going on the panel.

13:50
So as I said, we're going to be looking at this through the client journey.

13:54
So let's break this down into the four kind of basic segments.

13:58
There's a the sort of the clients seeking advice, they then go on to call an agency, the advice that then researches the problem.

14:08
And then we're thinking about how do we curate and manage that information and communicate that to the client.

14:13
So Dan, I'm going to start with you.

14:14
Let's set the scene here.

14:16
When we think about clients searching for advice, how might a client typically go about seeking advice online?

14:24
Thanks very much, Francis and Zach, great to be here.

14:28
Well, well that that is $1,000,000 question.

14:32
How might a client typically go about seeking information?

14:37
I think the honest answer is it varies a lot depending on the service that you're running.

14:45
So I'm really interested to see what people put in the the Q&A.

14:54
But my question here would be what's the community that you're advising?

14:59
Who's the community that you're advising?

15:02
What do you know about them?

15:06
How can you be curious about the journeys people have taken to get to you, to get to that point where they have contacted your service?

15:16
What do you know about the steps that they have typically taken before that?

15:22
That does vary quite a lot and I think the main point that I want to make here is that we are seeing quite rapid changes in people's behaviour, how they go about seeking that initial information.

15:39
So we are seeing new search behaviour.

15:43
I'm sure people on the webinar will have experienced that themselves.

15:51
If you type into Google Now a a particular question, very often it will present to you an AI summary of a response to that question.

16:07
It will be the experiences of people who might be view themselves as digital natives or people who might be digitally excluded could be very different.

16:22
People using phones, people with access to smartphones, people with access to laptops, desktop computers might have very, very different experiences.

16:35
People asking a friend or family member who they think will be able to help them and then that friend or family member going online and again taking that action to try and find some initial information.

16:55
Well, we can say a little bit more.

16:57
I think we'll probably get into that about the different stages of the advice journey, but very often at this point people are thinking about the things that matter to them most and we need to understand that they vary.

17:16
So could be convenience.

17:18
They may be looking for a quick and convenient way to get some initial information.

17:24
We shouldn't underestimate that.

17:26
They may want to speak to a human being.

17:30
They may be worried about rocking the boat and worried about what the consequences of getting advice or seeking help with their situation might be.

17:41
They might be particularly interested in anonymity, confidentiality, that they might not be there really through their own choice.

17:49
They might be being referred to a service because that there's an issue that a third party wants to resolve.

17:57
They might be, they might feel that they're there because their housing officer wants them there, that there, that that's a condition of getting help from a housing association or from a local authority.

18:11
So it does vary.

18:13
Be aware of what the new search behaviour and what is happening in terms of people's initial search approach.

18:22
That is a significant change.

18:26
Yeah, I think that's really interesting point to bring up of this.

18:29
Yeah, this.

18:29
We are actively seeing a difference in how people are searching and finding that information.

18:34
And Lindsay, I want to bring you in here because of course the Advice Service Alliance provides this advice quality standard, which obviously has this whole dedicated section to about access to service.

18:43
What are we generally expecting agencies to do in terms of making sure their community is aware of their services and signposting information?

18:52
Yes, thank you.

18:54
Well, in many ways, I think making sure that the, you know, the visibility is there for, for communities that you want to serve.

19:03
And I think Dell's very right in sort of saying we need to know an awful lot more about your, your, the people you're trying to provide a service to, to make sure that you are tailoring that service in a, in a way that would attract them.

19:20
I mean, it's going to be very difficult because there's going to be a lot of noise out there.

19:25
If people are trying to look for advice and they, they put a question into a search engine or, or they ask a question on, you know, something like GBT, there will be lots of noise, white noise coming, coming back to them.

19:44
And so making yourself distinguishable is going to be quite a challenge.

19:50
I mean, obviously we would suggest that people do make sure that their their websites are very clear about what it is that they do.

20:00
The better that the information is on your website, the more likely it is going to be to be picked up in the right sort of searches.

20:08
And and I know that some of the networks are indeed talking with some of the IT providers.

20:14
About making sure that we can get that those prompts that would send people towards advice services higher in in the returns.

20:27
I think also that's very important point that Dale made about, you know, the sort of the awareness and people knowing who you are and where you are actually means going out to communities a lot more.

20:44
We have to a certain extent been services that have just sort of said, oh, we're here, our doors are open, come and see us.

20:53
And some of that is driven by the fact that we have too many, too many clients for our our services to be able to cope with.

21:01
So we don't actually need to go out often and, and try and drum up more support or get closer to communities, but I think that will be increasingly important and, and how services start to adapt to this idea that we've, we will almost have a, a legal education role as well, talking to communities about maybe some of the questions that they should be asking.

21:28
So I think all of that is is going to be really important in thinking as as we move forward.

21:37
Yeah, and I think that's really interesting what you raised there of yeah, both you need to know your community, but also the fact that there is this sort of you do need to start thinking about.

21:45
They call it Geo and AEO, but you know, being generative AI and engine optimization and so forth.

21:51
But then also thinking deep back in the in the real world.

21:55
How do you actually go back out into, you know, and speak to people in person and making sure people know, you know, who and where you are?

22:02
Yes, and I, I think that is a point that I really want to emphasise is we, we have to really remember what our USP is.

22:10
What is it that people get from us that they really, really value?

22:14
And that personal interaction, I think and, and being very sensitive to the community and what their needs are are all part of that.

22:25
Baoji, I'd love to bring you in here because of course with Nucleus, are you finding clients are using AI or coming to you via things like AI summaries at all?

22:35
Yes, I mean, as Dell said, lots of our clients want face to face advice.

22:39
That's what we, that's what we specialised in because that's our client group talking about knowing your clients, lots of illiteracy issues like language issues and so on.

22:48
And we find that's our USP.

22:52
But increasingly we are getting people searching online initially.

22:56
And again, as Dell mentioned, Google ChatGPT, they're all designed around the AI experience.

23:02
They're pushing it more and more now.

23:05
They're designed to keep them on, on their platform and almost like there's a discussion going on with the search engine in yourselves.

23:11
So by the time they come to us, they've thought about it a bit.

23:14
Something we've always told, you know, we've always encouraged clients to do, do a bit of research, look, really run the written, run the subject before calling it sometime.

23:22
But what we are also finding is that some clients have gone down that route to quite a large extent.

23:28
So by the time they've come to us, they've really tried some solutions.

23:32
Some of them might have worked, they would work with certain client group, but not with all of our client groups.

23:37
And I think that's the problem because some of them come back to us with even more complex issues.

23:43
And our clients live, everyone lives a complex life, but our clients in particular have complicated needs, let's say and inter interrelated needs.

23:53
And that's why the advice sector, I think has been really good at developing areas where social welfare areas where they interact.

24:03
So you've got people, people who are a little likely to be in debt, like to have housing issues, they're likely to have benefit issues.

24:08
You like to be in poor, you know, of low wages and so on.

24:11
So they all do interact.

24:12
And a typical example of that is we've seen people coming in with ask about Dr OS and, and they've got lots of information and they've made the decision in their own head.

24:22
They've made the decision, this is the route I want to go down.

24:25
This is the route, this is it for me.

24:27
But when they come to us, we have to, in effect, rewind the interview process a little bit and say, have you, have you, have you looked at what impact this is going to have in your housing?

24:37
Because the DRO is going to have a massive impact.

24:39
I won't go too technical down that, down that route.

24:41
But you know, it does depend on your tenancy.

24:43
It depends on your landlord.

24:44
It depends on, for example, local authority policy, how they treat Dr OS how and what's going to happen once you do that deal.

24:51
You could be made homeless and how that's going to impact your homelessness.

24:54
And that's all interrelated and our advisors are experts in that in talking to the client, getting the whole picture.

25:02
And as Lizzie Lindsay said, we're about knowing our clients, knowing the community, no, and knowing the individual that's in front of us, you know, and exploring in a face to face interview.

25:13
Coming back to the question, which was about AI searches, yeah, what we get in a lot of we've noticed is subjects we're not doing, we get loads of emails about, and this is from social subscribers, client social subscribers, NHS staff, all kinds of housing offices, as as was mentioned earlier on, where someone's put all the questions in into ChatGPT.

25:37
And I've got a lovely essay asking lots and lots of questions, touching upon sort of, you know, healthy inequalities, all kinds of things, really, really complicated stuff.

25:46
But coming in through, through e-mail and immigration, it's been, it's been exploding.

25:51
We don't do immigration anymore, but we get loads of people because they're desperate.

25:55
They've got nowhere else to turn.

25:56
And that's all over the country.

25:58
So I think that's another thing we have to, we have to really look at the fact that because they're, so, the resources are so limited, people are going to prepare really complicated cases and put them to you.

26:10
And we feel obliged to at least provide them with an answer or point them in the right direction.

26:14
And when they get so complicated, it is hard to start dissecting them and giving them a, a decent answer.

26:22
And Baljit, just to pick up on that.

26:26
And, and the fact that, you know, again, you may get, you're getting lots of enquiries in areas where you're not directly providing a service.

26:34
I, I think it does provide a kind of an obvious challenge for agencies to be thinking actually this is an, this is also an opportunity for us to be thinking about the networks and partnerships that we've got locally.

26:50
You know, how are we connected with other advice services?

26:56
How easy is it for us to ensure that client journeys don't break down when we we may be need to say it's not us that's gonna be able to provide this service, it's this other organisation.

27:14
There are obviously some great examples of local advice networks in London, elsewhere and also tech solutions that are trying to bring that together.

27:32
And I think that that is actually also a wider challenge for the advice sector.

27:38
Again, Yeah, it wouldn't be that property, Dale.

27:41
I think when an e-mail comes in where we feel obliged, we have to triage it.

27:46
If we get, we still have to analyse it.

27:48
You still have to sort like, you know, sort it out and say what can be done by whom?

27:52
And because ChatGPT does this a lot, because you asked all those questions and you and, you know, there's a conversation, you end up with a very long letter e-mail with everything thrown in it.

28:05
And it's designed to be positive as well, doesn't it?

28:08
So always put the positives and shine on it.

28:10
So there's nothing, there's no risk to the no weaknesses.

28:13
It's all strengths and pushing the clients guys.

28:16
And that makes it very difficult to analyse effectively in triage effectively.

28:21
And actually on that point of triage, that's kind of segues nicely into the next part of the client journey.

28:25
Because I think what you've said, but what you're all saying now I think also will bring will definitely come back up again towards the end.

28:32
But yeah, let's think about the fact that when someone's contact the agency, we, you know, and we're thinking, you know, for a particular relation to people calling up, we are now seeing by a lot of interesting AI tools, things like Wiser Assist, which enables you to transcribe and create a case summary notes, as you know.

28:51
And it's a really great tool that could hopefully speed up that process and hopefully then subsequently speed up the entire process of triaging.

29:00
Dal.

29:00
I know that advice UK have been doing, have been working quite closely with wiser.

29:04
Could you tell us a bit more about things like the wiser assist tool and what it offers in terms of benefits and also maybe we could cover some of the risks as well?

29:14
I, I, I won't, I won't maybe say too much about the practicalities because I think you can kind of hear about it.

29:20
You can hear about it from someone who's tried it directly and, and I might like let you say something about that, Baljit.

29:26
But essentially what Wiser Assist is, is a transcription tool.

29:36
So we Advice UK wanted to make it possible for our members and others in the independent advice sector to have access to AI transcription technology.

29:52
We wanted a slightly better power dynamic than just a sort of a direct relationship with a with a massive multinational.

30:01
So we we've been working with Wiser essentially to develop a product for social welfare advice services that enables transcription to take place and for then summaries of those transcriptions to be made available.

30:26
And there is also, as part of that, an integration with Advice Pro, the case management tool that we make available.

30:38
It was particularly important to us that as a tool it was able to transcribe people speaking in different accents that might commonly be found in communities across the UK.

30:56
That was really important.

30:58
We also wanted the flexibility to be able to swap different models.

31:05
So if we found that one model was better than another that it was possible to swap to that better model and not be tied in again to a particular model and and a particular provider.

31:18
So it kind of gives a little bit more, we hope control to the advice sector and a bit more of a sort of a, the, the a better power relation, dare I say.

31:32
But Balji, I don't know if you want to say something directly about your experiences of making it work in practise.

31:39
Yeah, yeah, sure.

31:41
We, we, we've been using wise for about 18 months.

31:44
We were lucky enough we'd been spotted a, a grant and that's the only reason we, we thought about it because obviously there is a cost.

31:51
But 18 months ago it was the MAPS debt modernisation programme, I think it was yeah.

31:59
And initially we did play around with the sites into the script and we we I got together a group of people who were interested in it, who were keen.

32:08
And this this was mainly because some people were already using chat BGGPT to, you know, describe certain things.

32:16
And we were noticing people were using it.

32:17
So we were quite interested in seeing adopted some kind of script and see if we could we could help people.

32:25
So coming back to it, yeah, it's, we applied for it and we, as I said, we were successful.

32:35
I'm just sort of sure I won't try and give a summarised version because I can give you a very long version which I'm sure you that nobody wants to see here.

32:44
Initially when you look, you're looking at these separate, separate websites.

32:48
It was usable and we got some really good results out of it.

32:51
But I think where it really started transforming our device provision was when it was integrated into Advice Pro.

32:58
We'd happily use Advice Pro and that's when it had the biggest impact because until you do that, I think it's hard for advisors to go from one site to another site to another site.

33:11
And you do want to integrate because the whole point of the AI experience, I think is to try and reduce people's workloads.

33:19
It's to try and reduce the friction that people have going to lots of different sites and lots of different platforms.

33:24
And we work across a number of platforms.

33:27
Unfortunately, we use Salesforce as well.

33:30
So that's when it really started to work.

33:31
And the approach we took was, as I said, we, we started with a for the pilot, tried it with those people, created templates which were in line with our wave recording because that was important to us in line with our fire management policy, in line with AQA, the quality standards as well.

33:50
So we, that's what we wanted because we like, you know, this is what the client told us, this is the advice we give them and this is what happens next.

33:57
And that, and that's really good for, for quality.

34:00
So as, as we adapted the, as we developed the, the templates, it got better and better.

34:07
And I think that's when people really started noticing it because what we tried to do is have a really short feedback loop.

34:14
So as soon as somebody noticed an issue to report it back.

34:18
And this is where it was really good that we had a keen group, you know, of people who were involved in, in the pilot project because they wanted to feedback.

34:25
And that's what led to improving really, really quickly.

34:30
And the other thing which I think working with Wiser, which helped any company, but I think it's worth getting an expert company in, is that we were able to train it faster by talking to them, telling them what the issues were.

34:44
And they were able to speed up the learning process and, and enhance the bits we liked.

34:49
And, you know, minimise any, any of the hallucinations and all the other things that were coming up.

34:54
There's plenty of that at the beginning.

34:55
And, you know, people who were great laughs about some of the, some of the sentences that came out of it.

35:00
But very quickly working with Wiser, they were able to tune it to exactly what we wanted and, and the, and the conditions we were using it in.

35:09
So that was good.

35:11
And then we expanded to the rest of the organisation.

35:14
And I think that's when we, we realise how, how much we'd have to keep looking after the processes because it's not just throw it in there, fix it in and that's it.

35:26
It's a little engine that's going to keep going because we took it off some, some areas because people just weren't using it.

35:33
Other areas we extended to like the phones with the volunteers, that was really good.

35:38
That worked really well.

35:39
I'll come on back on to why that worked really well in a in a minute.

35:42
And we interestingly long interviews, especially debt interviews, debt and debt, housing and benefit, we had to develop different templates for each of those areas because people getting different information houses a lot more legislation based.

35:57
So they wanted that in the, in the notes debt a lot more figures and things like that and some, some some confidential information which we wanted excluded and things like that.

36:06
So it's a lot of tinkering around with the templates to make it to increase the quality of the thing.

36:12
I would say we could have got what we wanted quite quickly, but in order to actually improve it thereafter and use AI to actually improve the, the quality of the service even more, the templates I think were a really, really important part.

36:27
Coming back to the phones with the volunteers, what that did, because we, we've run a telephone line and we just want to put lots and lots of calls coming in.

36:38
So, so, and the advisors are answering as many calls as they can.

36:41
They, they, they make the notes and then move them to the next one.

36:46
The script allowed them to do a lot more calls, which was, yeah, suddenly the numbers shot up and which is a bit worrying actually, because it did.

36:56
You can lead to burnout because you can get through a lot more work on that.

37:01
So we had to change what we actually wanted from it as well.

37:06
Look at what we're asking for volunteers and say, you know, what we want to do is answer all the calls, not necessarily answer them in change the triage process we were talking about before.

37:17
So there's a lot of tinker you've got to change your procedures as well, I think.

37:21
And the great thing is we're, we're up at 8595% of call answer rate, which is great.

37:27
But we don't answer them all in great length because you can do that, but you, it can lead to burner.

37:32
The great thing was, I think from staff and volunteers point of view is 1.

37:37
I think it was quite exciting, but especially about 18 months ago to be involved with AI because it was everywhere.

37:43
Yeah.

37:43
And so people were quite interested in it naturally.

37:46
I think people are really, really happy that we were investing in the service as well because we went over the last 10, we cut off, cut off, cut.

37:54
It's all about doing more with less, that kind of thing.

37:57
But this, the way we introduced it, it was to say, let's look at the quality, let's look at what we can do better as opposed to let's just see if we can get through more work.

38:06
And I think that's got to be at the key.

38:08
That's I for me, that's got to be central as well, not just to use the spare capacity to do more work.

38:15
I think we've got to focus on what work we want to do and do that really well.

38:21
Yeah, and take some load off.

38:23
Sorry, you can say such.

38:25
No, exactly.

38:25
I was just going to say, I think it really comes back to this question of quality, right?

38:28
And making sure that it's the quality of work remains high even when we're integrating these new things and thinking about how it's, as you were saying, integrating all these different procedures and things, which Lindsay, I'd love to bring you in here because I think obviously, yeah, the big part of adoption of this is quality.

38:45
And and do you think these tools need new standards alongside adoption or would this just be reinventing the will in terms of what's available already to agencies?

38:55
I mean, I think I think it's really interesting.

38:57
And I've, you know, I really vow to belch its experience on on this and and his his long history of working in in an advice, but I'm absolutely willing to bet my house that this is not going to lead to less work.

39:13
I'm pretty sure that AI is going to generate in itself more work for us to do.

39:21
And there are some things that are a little bit worrying about that really.

39:25
But, and I think what we're seeing at the moment is as as the AI tools, they're a bit rough and ready, but they're going to develop and they're going to get better.

39:35
We know that they're going to the searches are going to produce more accurate results.

39:40
We will get better at the, you know, the, the transcriptions.

39:45
So all of that is going to improve and we will get get that immediate benefit.

39:52
So we will see immediately people start saying, oh, what a relief.

39:56
You know, we're, we're not having to spend hours writing up case notes, etcetera.

40:01
But that will fade, that will fade over over time.

40:06
So that's an immediate payback.

40:08
But as time goes on, we won't appreciate that quite so much.

40:13
But the things that we will need to consider are whether those notes are this what we need in order to be able to progress the, the cases and my and, and so for well, this just using this, the, the one example of, of being able to write, write up case notes.

40:34
One of the things that we, I would be really interested to find out more about is how those notes are reflected in supervision, for example, and particularly that kind of feedback loop with advisors, because you know, it's, it's very important way of understanding what the training needs of an advisor is.

40:56
If you look at the, the case notes and you think, oh, there's something really a bit dodgy here.

41:01
Now, if that's just a, a transcript that's been sort of polished up, is that going to have an impact on the, the kind of supervision and that, that sort of learning loop really?

41:15
And then also, and these are about very human aspects of giving advice is the question of whether you know, you've heard a call, you've you know, you've talked it all through with the client.

41:28
You come off the call or come out of the interview and then you write up your notes.

41:34
And in a way that's kind of firming up that conversation in your head.

41:40
So you you're thinking through what the advice was that you gave.

41:45
Did you explore everything and then you're sort of codifying it for yourself in those case notes?

41:53
And so there is a real worry.

41:55
Is that going to have an impact on the training of advisors, their understanding of the law?

42:04
And I think, you know, we're at such an early stage with it.

42:08
I think we really need to be thinking through how do we monitor some of that?

42:12
How do we make sure that we don't end up deskilling and and effectively losing all that kind of really in depth knowledge that we've got about how the law works in relation to people with with limited means?

42:30
Can I just come in there?

42:31
Lindsay, it's interesting you say that.

42:33
That's the way I work with the notes.

42:35
You work it out even in interviews, you sort of you scribble around and you work it out in your own daddy.

42:40
You get things sorted.

42:42
We have found.

42:43
Yeah, you don't need that because the notes have been taken.

42:47
It does change the nature of the interview because in the interview you've got a space to do that.

42:52
And I think looking at quality again, what we're trying to do, and I'm not sure we're succeeding yet, but trying to do is use the extra time that you'd make notes to have a more meaningful conversation with, with with the client.

43:04
Put the human back in the room.

43:06
The person being person centred.

43:08
It's hard to do that, but use the extra time, gain some of that time to make eye contact to, to talk about some of the issues and explore some of the issues and maybe explain the next steps to them.

43:22
Because I know some people are starting to.

43:23
It's like the old using the old dictaphone, you know, you'd change it to summarise it at the bottom at the end.

43:28
And hopefully what that has to do is, is say in the next steps and things like that.

43:32
So yeah, I think that will change over time, just the way we we hold interviews.

43:37
And I mean, it may be helpful again just for me to chip in here.

43:41
Certainly we've had some feedback from organisations using Wiser Assist where they're saying actually it's been really helpful as a way of really trying to understand what has happened in the interview.

43:54
And I think all of us who've had direct advice experience will recognise that sometimes our completed case notes can actually reflect a much more polished version of what actually happened then was the reality for whatever reason, often for good reasons.

44:18
You know, perhaps the, perhaps the client was really upset, Perhaps, you know, perhaps we had communication challenges that we were trying to overcome.

44:33
It's easy for for completed case notes not to be an accurate reflection of what really happened.

44:41
So they're a kind of there, there are, there are sort of two sides to that quality coin.

44:51
Sometimes that transcription can actually help you understand.

44:55
Oh, actually, no, you know, the way that the interview was carried out maybe was not as good as it could be or that there there were things that could be improved about that.

45:08
I do think it leads to more skilful interviews as well.

45:11
You can have more skilful interviews and people and I think as, as you mentioned, advisors can show this because you could see the difference between the skilled advisor having a conversation and the way it's, it's progressing and somebody who's doing it looking up and reflecting on it.

45:29
And, you know, at the early stages of the career.

45:31
Maybe I, I think that's really interesting about it because I think with these tools that, that are augmenting practise, I think what they can do is really help with people at the, at the beginning stage of their career, or maybe practitioners who don't quite get the processes as as well and are not getting, you know, quite getting the message over of what we're trying to do.

45:57
And I think they can pull that, that sort of level of advice up, which is, is really, really helpful.

46:05
I think the bit that worries me is at the other end of it, are they making us less skilled as, as practitioners?

46:15
And I can, I can think of all sorts of examples of a piece of work I was involved with in probation service where we saw good practitioners becoming ordinary practitioners when we, we brought in a new process.

46:29
And, and you know, there's, there's other examples as well, which, you know, we, we could pull on.

46:34
So it's, it's thinking about in the longer term, how these human processes will be could be if we're not really, really careful, because it's about thinking about it and, and trying to make sure that we don't lose some of that.

46:49
But those those, the thinking part of it, the understanding, the creativity part of what advice giving is about doesn't get lost.

47:01
I think on that.

47:01
Then Zuri, it's yeah, I was going to just because I think this really relates closely to what we were going to discuss next regarding kind of seeking information.

47:10
Because of course the specialised chat bots that are coming up, there's this exact same concern, right?

47:16
It's this question of deskilling and the role of agency and all these concerns that you've mentioned.

47:22
So you know, we know that you know, these specialised bots, things like assistance advice bureaus, Caddy or debt free advices Moneypenny are there and designed to help advisors seek out the right information they need to support a client.

47:40
So unlike general tools like ChatGPT, just to kind of help distinguish them for, for the audience, these specialised tools use a thing called a rag system.

47:51
So that recurs to a retrieval augmented generation.

47:55
Basically what that means is they restrict what resources the AI system can access and use as part of resources.

48:01
So ChatGPT, if you ask the same question, would go out onto the web and do a simple web search.

48:06
Well, it's a bit more than a simple, but you know, it's designed that it's just looking online.

48:10
This would be about restricting it, but the key issue here is that inputting this information can help us design more accurate systems, but doing that without permission is obviously an infringement of copyright and legal terms and conditions.

48:23
And us, we at CPG have obviously had this as a major concern, which is why we're actually partnering with people like Citizens Advice Bureau to support Caddy and debt free advices Moneypenny to ensure that these tools are drawing from expert and trusted sources to provide that high quality advice and the correct patients in place.

48:41
But then there is the store, this question of like making sure people are using it effectively to maximise the use of it and not just giving away their abilities and their skills and their expertise by simply just sort of offloading it to the chat bots.

48:58
So yeah, Lindsay, perhaps I'm staying with you.

49:00
You know, again, perhaps on this question of quality, again, what you know, the things you've mentioned already, would you like to discuss a little bit more in terms of what these chat bots might offer and then also what we should be aware of as they become more widely adopted?

49:14
Yeah, I mean, there are obviously there are some some issues in in, in quality.

49:20
And I think I think you've described that very, very well.

49:23
Really.

49:24
But you know that that's the idea that the just because you've got information doesn't mean to say you've got knowledge and having information at your fingertips.

49:36
It needs to be processed and put into a kind of framework of understanding and, and, and learning really.

49:43
And that's both for the advisor, but also for, for the public.

49:49
And I think it kind of for me, it feeds in not only to that kind of quality issue.

49:56
And, and, you know, we, we know things like help lines are going to be really very, very challenged in the future because a lot of the things that people would phone up help lines for are, are questions that, that AI will be able to, to, to help, to answer.

50:15
But there is, you know, there is this concern then about where is that information being generated from in, in the 1st place And, and who's paying for that?

50:26
Because the, the, the risk going forward is that funders start to think, you know, oh, well, that's, you know, people can get the information through through that wonderful AI kind of driven helpline.

50:43
We don't need advice services anymore.

50:45
We don't need people to, to kind of meet face to face.

50:49
It's very expensive, you know, so there there are some real challenges there, not just about, you know, about thinking who who is likely to be paying for that and how do we keep that going.

51:04
And so I'm thinking about things like, you know, the websites which are generating the information.

51:10
So for, you know, child Poverty Action group websites, fantastic.

51:14
And I'm often, you know, suggesting that people look at that others like advice now, Citizens Advice, they're all they're all, you know, fantastic sources, but they need people with expertise to to sort of populate those.

51:31
And, you know, thinking right back to the early days that I was started working in advice, you know, it was the skilled practitioners who were wading often not not necessarily legally trained, but the people who were wading through legislation to, you know, the acts and then the regulation.

51:52
And then the guidance that came out that was starting to spot all those kind of contradictions or the ways in which, you know, policies can be implemented.

52:04
And that's the sort of information that you get through, you know, a sort of skilled practitioner website that you wouldn't necessarily be able to get hold of otherwise.

52:15
And so there is a there's there's quite a broad almost existential kind of question going on here.

52:24
And I think I'll come back again to that.

52:26
Let's let's be really clear about what we're really good at and make sure that the funders and the government generally know what we're good at and that we really kind of promote ourselves and and also the audience that we're we're really good at servicing.

52:43
Because I think again, most people chat box and things like that.

52:47
They work for the general public and for most queries they'd work the vast majority.

52:53
But the clients who end up with advice agencies tend to either have used it and ended up with more issues or have complex issues that are none other interrelated.

53:03
So I think it does.

53:04
And that's where really skilled advisers are able to unpick it and with their connections with other people, with other, you know, with say people in, in different teams, different organisations, different partnerships bringing in those things.

53:18
And I think there's a real opportunity here that people are becoming more local.

53:22
If you look at the crisis, crisis and resilience funds that are coming out, that's all about focusing in the most poor, isn't it?

53:28
When people have got multiple issues and it's about working the the the not-for-profit sector, working with local authorities and local authorities finally agreeing that there's there's a role for lots of community organisation to support the local community.

53:45
Yeah, I don't think so we're ever going to meet that need.

53:48
Yeah.

53:48
And I think kind kind of what was said earlier is that, yeah, it's about making sure these tools are used by experts.

53:55
And I think what I'd love to raise was you will now and perhaps direct you to this, to Dell, is the question of training and the role of supervisors in all of this.

54:05
Because yes, advisors might use these specialised checkpoints, but how do we make sure that if you're earlier in your career, you still have that expertise and that confidence to make sure that advice is correct and can be trusted?

54:21
It's a really good question.

54:23
You know, I think there is at a very basic level.

54:28
So first things, if you're, if you are currently in a, a supervisory role, you need time for yourself and for your own professional development to think about how these tools are going to impact your, your practise.

54:45
You, you need time to reflect on that.

54:48
I think you need time to be able to experiment, to try things out in a Safeway, to be able to compare notes maybe with peers in other services.

55:03
You know, Belgium, it's, it's great hearing about, you know, again, how you've created the space within Nucleus for people to be able to yeah, you know, to, to, to think about what, what works.

55:21
We, I, I think people may have seen it in, in the chat, I've put a link there to a framework that we use for the sixth stage advice process, which, which again, may be helpful.

55:37
So I I think increasingly supervisors will need to be able to manage advisors who are working with multiple different sources of information.

55:51
You'll need to be able to give people advice and support on how to interrogate outputs of of answer engines of different generative and Geo search.

56:10
For example.

56:12
I I think again, and and this is probably reflected in the level of complex casework that people carry and the fact that there are problems that are interrelated.

56:23
And supervisors will need to have more experience and more knowledge about how to combine expertise in different areas of social welfare law and particularly in those common cluster areas that we might come across.

56:44
And I, I think again, that will inevitably become more important.

56:49
There'll be there'll be a greater need for people to be able to, to supervise when it's when you're not just as it were working out the technical advice in a narrow area of social welfare law.

57:04
But actually how you're going to combine that together and how you're going to combine it with people getting the support they need to access on the advice and make progress.

57:19
And again, maybe the, the briefing, I think it's also gone in the chat about better understanding of client journeys end to end.

57:28
I think that knowledge will become increasingly important for supervisors.

57:33
How do we make sure that the client's journey end to end is working well and not just a single advice transaction.

57:43
What what's what's the outcome of a particular piece of work?

57:46
Now that's great.

57:49
And I think that kind of leads us.

57:50
We've got sort of just a few minutes left of the panel.

57:53
So to moving to that sort of last segment, so to speak, of thinking about managing that information and communicating with the client.

58:01
We've mentioned a few things already around this, but I think there's one thing here that perhaps we haven't talked about that I think is probably worth discussing.

58:09
And that's questions around how do we communicate our use of tools to clients and whether that's something we should be.

58:15
And if so, how do we do that?

58:17
And how do we manage the responses that might come out of that, both positive and negative, if any of the panellists would like to start off, I mean, I can just kick off very, very briefly in terms of data governance.

58:32
So, so I'm sure you know, all of our members, certainly big, big things that they needed to address with various changes to GDPRU, KG, DPR and, and I think that that data governance is just, it reinforces the importance of getting that right and making sure that you are thinking that you've got the right data governance in place and that members of your team understand what what that looks like in practise.

59:11
Again, I think that focus not being obsessed by just thinking about consent in relation to data governance, but thinking about how you communicate with clients throughout their their journey so that they understand and it's easy for them to find out how you're processing their data.

59:36
And that they're not surprised to find out that you were processing their data in a particular way and coming back to that, those key steps of the journey, Trust, so important building trust, showing that you're a trustworthy partner with people's data.

59:58
And obviously given the sensitivity of the data that we work with.

1:00:02
Yeah.

1:00:09
Can I chip in here and mention all kind of thoughts that we're having about this in relation to the advice quality standard?

1:00:20
Because we are thinking that one of the things that organisations are going to really urgently need to be able to demonstrate is that they have a policy in relation to their use of AI.

1:00:37
We have at very, very early stages thinking about that at the moment, but it's very clear that it will need to include quite a few sort of considerations really.

1:00:50
So I mean, the ethics of it all I think is, is really important that we we grapple with that.

1:00:57
But issues around security, you know, are you using AI in a way that protects your own data security and protects, protect, protects the confidentiality that we want to give to clients?

1:01:15
And we, you know, we know already that there's been issues around that, but we, we really do need people to kind of keep on top of it really because as AI changes, then it's likely that that that will change as well.

1:01:29
But there's also issues about transparency, you know, are we being open with clients about where we're using it and how we're using it?

1:01:41
But also thinking about issues in relation to insurance.

1:01:44
So for example, if we've used it and we've given the wrong advice, how does that all then pan out in relation to who is responsible for that?

1:01:54
And where do advice organisations sit?

1:01:58
And then that point Dal that you've just mentioned about trust, I think there's another issue there which is about authenticity.

1:02:08
You know, how is it that we, we communicate to clients that we are, we are, we are authentic human beings, that we have genuine compassion and that we, we show that in our, in our communication with them.

1:02:25
And you know the, I mean, again, maybe it will improve, but at the moment you can often feel that something that's been AI generated is, is not authentic.

1:02:38
And I think we need to try and hang on to that as, as as much as we can.

1:02:48
And Balji, I know that you, you mentioned earlier that you were use, you have an AI policy in place as well.

1:02:54
Tell us a little bit more about how you've been communicating your use of AI to to your clients.

1:03:01
It's a bit basic for as far as Dale was concerned.

1:03:03
It is having notices everywhere in the office.

1:03:07
It is having messages on the phones and so on to say we're recording it and we're using where we're very keen to make sure that people know that we want to analyse it to improve the service.

1:03:21
Obviously we've, and we're very keen as well to confirm that it gets deleted after a certain amount of time.

1:03:29
So as well as far as the phones are concerned, as far as any recordings are concerned, where they're stored, where they're not stored and how quickly they get, they get deleted.

1:03:37
And, and the very last thing I'll say on that is that we, the other thing we've been focusing on is that the case notes are the legal notes.

1:03:43
As far as we're concerned, it comes back to the human in the loop.

1:03:46
It's your decision.

1:03:47
You decide what's going to be edited out like general chat and things like that.

1:03:51
What additional you know that's the you don't get finished notes you get you get a very good script, but you add to it, you you adjust it and you sign it off and it'd be back to the supervisor signing it off.

1:04:04
Really lots of clients have what have have have said.

1:04:08
I don't want to record it.

1:04:09
I don't want to put through AI, which is fair enough when we would turn it off.

1:04:14
Can I, can I ask, do you do you ask clients about their use of AI phone calls?

1:04:21
No, we tell them and then they can opt out if need be.

1:04:25
And no one's and no one's decided to opt out the phones at the moment.

1:04:29
I think that technically we'd have to get in touch and try and get back in touch with them because it's a slightly more evolved process.

1:04:35
But yeah, face to face interviews.

1:04:36
Absolutely.

1:04:38
I just wondered whether when the client's been using AI or if the, you know, do you ask them, you know, is this is this letter you've sent to me a draft?

1:04:50
Have you generated that through AI or something, something along those lines?

1:04:55
We've never chatted, challenged it, but I think with some glances, obviously the clients who tend to have used it more than others have had language issues, English as a second language, things like that.

1:05:06
So it's very obvious when you meet them compared to the, the letter, the, the emails you've had from them.

1:05:13
Yeah.

1:05:13
But we've had a few people put things through AI.

1:05:16
We had a few volunteers doing that and just, you know, outrageous behaviour.

1:05:19
Some of the stuff we had to really sort that stamp on that.

1:05:21
That's, that's at the early stages, 18 months ago, you know, when people putting it through, they didn't think anything of it, you know, partly because it's, it's open and it's, you know, it's, it's not secure.

1:05:32
That was an issue.

1:05:33
But also the fact that you were talking what you were talking about earlier.

1:05:35
We want to know what you know and we want it to go through wiser so we know when you've put a used AI and what you've used and what you haven't do so we can make a proper assessment.

1:05:45
But that I think that's the real thing.

1:05:48
Everyone's using it at the moment and that's what we found.

1:05:52
I was so surprised about a year ago that every other people certainly were using it and clients and advisors and it's the norm, especially for young people.

1:06:00
It's the norm.

1:06:02
Yeah.

1:06:02
And I suppose the challenge that it kind of goes back to what Lindsay was saying about this question of accountability and then the impact that has on things like insurance and the practicalities of this, right?

1:06:13
Yeah, I would.

1:06:16
I think it comes back to it is the authenticity and treating people like real people.

1:06:21
And we want, we want to keep that relationship going.

1:06:23
That's that, that is the most important relationship we've got with the client, isn't it?

1:06:27
The fact that it is totally confidential.

1:06:30
It is about them and we will look at the whole situation and get and assess it rather than just reply to a question, which is what AI is doing.

1:06:40
You know, I, I, I, I think I'd add kind of, you know, for, for our members, for for example, our members from DDPOS, from other user LED organisations.

1:06:58
I think there is, there are, there are definite, there are definite questions about how we can make sure again, that we retain agency as communities, as community advice services, that, that we can, that we're connected with those wider drivers of we're, we're, you know, I'm sure people are very conscious just about challenges posed by misinformation, by content that that isn't trustworthy or, or that distorts or, or that's trying to scam people.

1:07:57
And I, and I think again, that's, that's for me, one of them huge strengths of, of independent community based advice that it, that there are opportunities for people to come together and again, build their agency in identifying the problems, the issues that affect them.

1:08:20
And also thinking about some wider systems change.

1:08:26
And, you know, again, we haven't got time to kind of go into some of those wider power dynamics.

1:08:34
But I know that that again, is a significant issue for for many people who are thinking about how AI is being used and its impact on, on civil society.

1:08:45
Yeah, great.

1:08:47
Well, thank you all.

1:08:49
I can see there's loads of questions coming through.

1:08:51
So I want to make sure we have lots of time to to try and get as many through as through as many as we can.

1:08:58
So what I'll do is I'll start off with, we had a couple of few through the inbox coming up in the week before.

1:09:05
So one person asks, how might advisors actually kind of quite connected to what we've just talking about, how my advisors differentiate themselves from AI and what do advisors need to show to clients that they are reliable as AI or even have greater reliability?

1:09:22
So I think maybe comes back to this question of trust, right?

1:09:24
What you're just discussing.

1:09:27
I, I think so is that, I think that's always been the issue in some ways.

1:09:31
How do we get the trust of a client very quickly and face to 1 is being a trusted, you know, organisation on the on the street of the local organisation.

1:09:40
That's one thing, branding and all that.

1:09:43
But at the end of the day, it, it comes down to the advisor, the face to face interview.

1:09:48
What's the approach, what's the values, the organisation?

1:09:52
How do you treat an individual when they come in?

1:09:54
How do you treat the individual throughout the interview?

1:09:57
And as we said before, switch, switch the recorded, if they don't want it, treat them as a human being, treat them as a, as a, as a person.

1:10:04
And I think the other thing is coming back to, I think some of the services are becoming hyper hyperlocal.

1:10:09
Some there's a lot of discretion.

1:10:10
Local authorities.

1:10:11
We know that better than other because we've been doing it all the time in which everywhere you're working in, the more knowledge you've got that comes out and into you.

1:10:19
That's the first thing.

1:10:20
And secondly, if you can create a decent case plan at the beginning that works wonders with a client, you can see the stress literally drop off an individual when you're talking to them and you can see them sort of like trusting you.

1:10:34
And you know, and this comes back to what we were saying before, skilled advisors compared to people sort of starting off or building their skills, you can see in different interviews that that trust comes about a lot quicker.

1:10:43
And I think AI, it is going to augment it augments lots of different aspects of it is how we use it, isn't it really?

1:10:51
And I think this is another reason why we weren't, you know, we were pleased with having a product such as wires and advice by working together, we didn't have to think about it.

1:11:00
We could put it in a little box.

1:11:02
It will do the magic right.

1:11:04
And hopefully they'll come up with, with other, other, other ideas that we can, we can, we can use.

1:11:09
But the bottom line is we want to provide advice.

1:11:10
That's what we're really about.

1:11:13
And we are about providing advice to our local communities.

1:11:16
So I think it comes back to it comes back to the original aims and objectives of the of the charity.

1:11:23
I mean, I've, I've popped something in the chat, which is just some, some of our top tips around a whole person approach and, and just some practical suggestions the services can take.

1:11:39
I think it's, it's fair again, just to go back and remind ourselves that at the moment, AI can come across as, as, as Baojit said, as as being, you know, very kind of, you know, polished and, and, and positive.

1:11:56
It's still remarkably difficult for to safely have some sort of AI product that really understands someone's journey from end to end and what's important to them.

1:12:15
So, you know, again, many transcription tools, they're just working on that transcription.

1:12:21
They don't have the previous interactions and they're not referencing those previous transactions in the way that they transcribe.

1:12:33
And for most tools at any rate, not not in the way that they are producing summaries.

1:12:40
So again, you know, for us it's that key point of difference is you're human, you empathise with people, you understand and you can show that you understand what the journey looks like from start to finish.

1:12:58
And also what's really important to the client, what, what, what, what's the outcome that they would really like to see and what are realistic ways of getting there?

1:13:10
Yeah, not, not just a sort of, you know, a hopefully optimistic way of getting there.

1:13:16
Yeah.

1:13:17
And I think we're in danger of completely agreeing on absolutely everything here.

1:13:25
I mean, I think there is, you know, how how we actually demonstrate that is, is really important.

1:13:32
And I have I felt when I rejoined the advice sector about 10 years ago that we, we were very much focused on service delivery and thinking about how many people we saw.

1:13:50
And that was the commissioning model that was those popular from government and funders at that point.

1:13:58
And in a way we had lost something that we'd gained stuff.

1:14:02
We were much bigger sector than we were when I, I'd first joined.

1:14:06
But we'd lost a lot of that kind of community activism, that kind of understanding, you know, how how government policies are actually being implemented and then lobbying on behalf of of organisations.

1:14:22
And I think, you know, some of that was the changes for those of us who are charities, the changes that we saw in the Charity Commission and how the Charity Commission was operating and its relationship with government contracts, etcetera.

1:14:36
If we're moving away from government funding, then that does free us up in a way to then think about, well, how do we use our collective experience to be able to not only to, you know, sort of challenge policies and, and implementation that may not be working on people's behalf, but also to go out to communities and to educate them about their rights And educate sounds the rather loaded term.

1:15:05
But you know, nevertheless is to make people aware of the fact that there are, you know, there's a legal framework and that they have human rights that operate within in that.

1:15:17
And I think the some of the benefits to that is that it will also play into strengthening people's feelings about the democratic process and where they where they fit in there as well.

1:15:29
So I think there's enormous opportunities that we can we can take if if we, if we so choose to.

1:15:37
Brilliant.

1:15:38
Thank you all.

1:15:40
Sorry I was on mute for a bit.

1:15:41
Another opportunity I think we were because we, this is all data LED.

1:15:45
So all this is collecting data.

1:15:47
We've been giving it away rather than giving away to chat, GPG or one big commercial.

1:15:52
It is an opportunity for us to analyse that data.

1:15:55
But you're saying, do we tell clients and work out what works and what helps tackle poverty and what doesn't?

1:16:03
I think that's a huge opportunity if we if we work together as a sector and, and, and get back into the campaigning mode.

1:16:09
Lindsey.

1:16:12
So I think there's an opportunity.

1:16:14
No brilliant, no love it.

1:16:17
So another question.

1:16:18
So we got another question from the inbox saying attention exists between the desire to innovate and improve and the need to address both staff and volunteers fears of change regarding the new technology and the risks of using AI in advice without properly considered approach.

1:16:34
What approach should be taken by management and the advice sector to introduce AI in the way we work?

1:16:40
And Baudi, I feel like this is probably one that's directed to you having a direct experience with this.

1:16:47
I think it is cautious and we haven't got all the answers.

1:16:50
It is, you know, you learn through the process, the process of change.

1:16:53
It is introducing it slowly as a pilot and listening to people like we do.

1:16:58
It's like any other policy.

1:16:59
It's no different than introducing, say data.

1:17:02
We introduced case management systems.

1:17:03
There was, there were similar issues around that, slightly different, different scale and so on.

1:17:09
But it is the same thing.

1:17:11
You've got to listen to the people on the front line as to how it will impact and and you work together to make sure you're coming back to what is, why are we here?

1:17:21
What is the point of the charity?

1:17:23
It's to deliver advice free on social welfare law and to tackle poverty.

1:17:27
So as long as we can show that I think we will take people with us.

1:17:31
It is about gaining different skills.

1:17:34
There are issues around using work and things like that or stacking work up to increasing the targets.

1:17:40
Again, that's what I'm service delivery and Lindsay touched upon that.

1:17:46
But I think it's natural within the sector.

1:17:49
But aside the funding issues and so on, it is natural to to look at quality and to look for long term solutions.

1:17:57
So if we put in that central and try and try and work that into the change, I think that that's the best we can hope for and move, move forward together.

1:18:08
Sorry I haven't got a complete answer because I don't think there is one.

1:18:11
It's difficult.

1:18:11
It's a challenging 1, isn't it?

1:18:13
Dow and Lindsay, would you like to add anything?

1:18:16
And I think the only thing that I think this is a slightly different problem we've got in a way as a sector is that people want to engage with it too quickly.

1:18:27
And, you know, oh, yeah, we'll have some of that, you know, isn't that?

1:18:30
And I think that we've got to be cautious about as well.

1:18:33
So it is.

1:18:34
It's about thinking about it.

1:18:36
Really, it all comes down to, you know, thinking it through very carefully, management, taking a strategic approach to it, making sure that if you have a board of trustees, that they're aware of what's happening and what's going on.

1:18:51
And that that, as Bausch is saying, you're trying to bring the team with you.

1:18:56
And you know, the chances are that people will be engaging with AI in their ordinary lives.

1:19:02
So you know that it's just there and it's going to be rolling out faster and faster.

1:19:08
So I think in that way it, it helps, but you know, making sure that people really understand that you've thought about it.

1:19:16
I think it's really important, you know, I completely agree.

1:19:23
I, I think the obvious way that you can show people in your, in your team, staff and volunteers that you value them is, is thinking about how you can invest in their skills and their ability to navigate this kind of world.

1:19:40
You might want to think again about, you know, in the same way that you might think about access to 2nd tier advice.

1:19:51
And that's an investment showing that you care about them and, and the needs that they may have for support.

1:19:59
Just thinking what other options there are for you to help people learn.

1:20:07
To gain knowledge that can I think address maybe unfounded fears and it can also probably promote some really good conversations where maybe actually, yeah, there are real, real fears which are well grounded.

1:20:23
You know, we've talked about some of them deskilling things like that.

1:20:29
Again, I'll just give to the the seat pack team.

1:20:32
There's people may have come across the Open University Gen AI course.

1:20:40
It's a free course.

1:20:42
So that's something that's accessible that you know, can be, can be shared.

1:20:49
Certainly advice UK, we have a, an AI briefing that you can, you know, you can signpost members of your team to, or look up yourself.

1:21:02
And you know, I, I know all of the major advice networks will again have will have similar resources and support that's available, but investing in your team to be able to respond to that change and, and giving them some space to think about it, to try things out.

1:21:25
Two very practical things, I think.

1:21:28
Thank you.

1:21:29
So I can see we've got lots of questions in the Q&A about risks of wrong info, which we did talk about a bit.

1:21:36
But there's an interesting question here about they say there's a major question unanswered about whether anyone or anybody can be held responsible for the output from AI systems.

1:21:46
Without responsibility, it'll be very difficult to hold anyone to account for the quality advice and information or even more importantly correct or output and follow up to update with corrected information.

1:21:58
And this includes the issues around evidencing and evidencing the basis and of what's being produced.

1:22:04
So I think, yeah, go back to this question of yeah, accountability, open to anyone that would like to respond.

1:22:11
I think it lies with ultimately with, with the organisation, but it's down to the, the human in the loop.

1:22:17
Again, they're signing things off.

1:22:19
And it is, it is terrible that we're putting all the pressure back on the advisor.

1:22:25
But I, I think, I don't think there's any other options.

1:22:27
They're the one in, in the interview, they're the skilled individual.

1:22:31
And it comes back to what we've been saying to it throughout this webinar.

1:22:35
I think this requires more, more skilling up and it's about getting capable people who know what they're doing up and run.

1:22:42
You know, you can't just, you know, low.

1:22:46
It's this shouldn't be about deskilling because at the end of the day, they're the person who's making this in signing it off and the organisation obviously is, is responsible insurance wise and so on.

1:22:57
But the responsibilities on the advisor and the responsibility on us to train the advisor to make sure they're able to do the job.

1:23:06
And I wonder whether connected to that, I mean, whether we could almost think about the advice supply chain, almost, you know, this way that all this information is getting to connected and whether there's an argument, you know, whether it's just creating better audit trails, you know, and that's even more important now that we incorporate different tools.

1:23:25
And that's just not, not just AI, but all these different tools that we're using in our day-to-day and making sure that we know who's, you know, accountable for what.

1:23:35
I just wanted to say I totally missed off the fact that I think using closed systems, that's where I think we need lots of new development because at the moment, I think the Internet's all about information.

1:23:44
It's not about quality data.

1:23:46
And we need to get back to having access to closed systems where you can, you can verify the information.

1:23:53
And it's, it's, it's written by professionals really that CPG lag and so on.

1:24:00
It's got to be, isn't it?

1:24:03
That's how you protect yourself and your clients.

1:24:06
You know, I mean, again, just to echo what Balgit has said in terms of insurance.

1:24:10
So we obviously provide professional indemnity insurance.

1:24:13
In some ways it's kind of quite simple, sort of nothing, nothing fundamentally has changed.

1:24:19
If someone thinks that the advice or information is coming from your organisation, you're responsible under the terms of your professional indemnity insurance.

1:24:32
We haven't thus far had a huge number of claims that relate to people saying I went on to this Advice UK's members website and it gave me really bad information and I'm and I'm now and I'm now suing them, which is great.

1:24:54
We're we're pleased about that.

1:24:56
We're obviously monitoring those claims closely.

1:25:01
But fundamentally, if there's something you host, you need a way of making sure that if it's giving advice and if someone would perceive that information as advice and not information that they need to apply themselves and reference themselves, you've got to be sure that that is correct.

1:25:22
I mean, again, you know, people may have come across some things like the Ava from access, from access social care.

1:25:34
Again, I'll pop pop a link in the chat.

1:25:37
That's, I think a good example of the, the retrieval augmented generation where again, the chat bot will be quite limited in what it can and, and the, the, the guard rails as to what it can and can't say and how it addresses that.

1:25:56
And you know, again, with any chat bot, you, you really need a way that if someone might get in touch with you and say, I've been given bad advice with this chat bot, you need a way of being able to audit.

1:26:11
Well, what, what advice did they get from it?

1:26:15
Is there a session there that we can identify if you're asking people to, to, to share information to when they use a chat bot that might give you some sort of audit tool, but you've, you've got to think through those issues of risk and liability.

1:26:32
Yeah.

1:26:34
And just to, just to add and come back to a point that you made, France is right at the beginning really, which is about, you know, how things are improving and learning.

1:26:45
And we, I think we all of us need to take responsibility for ensuring that the, the searches that are, are happening and the, the, the information that's being fed into the AI is as good as it possibly can be.

1:27:03
Because I'm sure that's the way that as, as these tools progress, but they'll become much, much more accurate.

1:27:11
And so the kind of risks that are attached should diminish.

1:27:18
I mean, we, we, you know, a lot of the cases are very well known where they've, it's thrown up information from wrong jurisdictions or from the past or, and, and I'm sure a lot of that stuff will start to start to disappear as we go along.

1:27:32
So that's, that's me with my optimistic hat on.

1:27:35
We love a bit of optimism.

1:27:38
It still does make it up though, if it if it doesn't know the answer.

1:27:41
AI is designed to make things up.

1:27:44
It's designed to be helpful.

1:27:45
That's its problem, I think.

1:27:47
Yeah, exactly.

1:27:48
And I think one of the major issues that we do have with bots is that people don't quite understand how generative AI works.

1:27:53
That as you say, it's, it's not designed to be, to understand what's true and not true.

1:27:59
It's understand, it's aiming to create a sort of human response, something that's creative and sort of random in the same way that talking to a human might feel.

1:28:08
And that's why you need to design, you know, a really structured prompts, you know, don't hallucinate, don't say if you don't know what the answer is to say you don't know because these are, you know, they're sycophantic.

1:28:20
They will just produce something thinking that you want it, and I think that can be a major issue if you're not sort of fully understanding of how it's working.

1:28:28
Underneath we've got lots of questions about the future of the sector, which might be a nice way to finish this panel, so let's have a look here.

1:28:40
How do we work actively and collaboratively with AI to help us learn and improve rather than getting it to just do something for us to cut down the task at the time?

1:28:50
And they say I'm interested in how AI can skill us rather than passively reducing our workload.

1:28:58
Open to the floor if anyone would like to have a go.

1:29:00
Well, shall I kick off?

1:29:07
Sorry.

1:29:07
Yeah.

1:29:07
And I, I'll.

1:29:08
I'll follow you.

1:29:10
Yeah, I'm, I mean, you know, all of these things, opportunities and, and threats really and the opportunities that that AI gives us is that those people who can self help will self help.

1:29:27
And, you know, if we think about the experience of the sort of lockdown, what we found, there was a lot of people who otherwise wouldn't have been seeking advice because the situation they found themselves in and because of the time that they had on their hands were finding their way to advice services where they would have otherwise found a different route to resolve their issues.

1:29:52
And and that in a way kind of somebody described it as elbowing out.

1:29:59
They're the kind of people who really are the ones who need our help where they may not have to have, they have very limited other options.

1:30:07
They may have issues around exclusion that are preventing them from getting the help that they need.

1:30:14
So I think from that point of view, it gives us the opportunity to kind of really focus on those people who we need to, we need to be, yeah, it really needs to be helping.

1:30:28
So I think I'll leave it there actually.

1:30:30
And yeah, I mean, there are some things which I think are probably covered in in some of the the the briefings that I shared.

1:30:45
But I do think one of the most powerful things that you can do is just try and carve out some time, even if it's like an hour, a couple of hours a morning, and just observe your current service in operation.

1:31:09
Just observe the service in operation and see what happens.

1:31:15
Make space to pay attention, to listen, to watch, to observe real people trying to navigate your service.

1:31:29
Start with that.

1:31:30
Start with their experiences with the things that matter to them.

1:31:35
And then maybe think about some of the real opportunities that there are to apply AI to those issues and to the to the challenges that people are facing.

1:31:50
You know, I think, and I think for me that means that rather than the technology wagging the dog as it were, the our purpose as organisations that are there to to improve people's lives, that that actually comes through and it helps to put things in the right order.

1:32:13
You know, it's a wonderful technical thing that calls can be rooted to the right person in big organisations.

1:32:22
But I think many of us might think, you know what, my experience of call centres has been pretty terrible.

1:32:30
So again, it's not necessarily just the technology, but it's actually how it's applied to a process.

1:32:39
And, and I think again, the best way of doing that is just taking some time out, looking at the process, looking at the things that are important to people.

1:32:49
And then maybe think of the tools, you know, if people want a real life example, went to an interesting event that debt free advice ran in London, You can see something about how they've tried to do it.

1:33:09
I think that works for their model.

1:33:12
Your, your service will be different.

1:33:15
But you might be thinking, yeah, what would what would that kind of energy and innovation look like in in our context, in the community that we work with?

1:33:25
So yeah, yeah.

1:33:31
On, on the link that they've shared, if you, if you Scroll down, they've they've got something there that they describe as advisor intelligence and not artificial intelligence.

1:33:46
So AI can be a way of building advisor intelligence, which is a nice way of putting it, I think.

1:33:52
Pithy.

1:33:54
Yeah.

1:33:55
Advisor first, tech second, right?

1:33:59
Brilliant.

1:34:00
Well, I think that's coming to the end of our time here today.

1:34:04
But let me just share my screen again because although this is the end of the webinar, this of course isn't the end of the conversation.

1:34:17
So, and I know Lindsay, you had a shout that you'd like to make regarding the AQS and consultation.

1:34:27
Yes, thank you for that.

1:34:28
Yes, as I, as I mentioned the AQS, we are actually in the process of reviewing the AQS at the moment.

1:34:37
So it's very, very timely for us to be giving some thought to all of this and giving some thought to how we strengthen the AQS to ensure that we really are measuring the things that really do get at quality.

1:34:52
And also that we are we have the right processes and policies in place that protect clients as well and reflects the kind of technology that's out there.

1:35:03
So if anybody does have any thoughts about the advice quality standard, please do contact me.

1:35:11
I'm really happy to put you in touch with my team.

1:35:16
We'll be taking as many views into account as as we move forward with this.

1:35:21
So thank you.

1:35:23
Thanks very much, Francis for that opportunity.

1:35:26
Well, you're welcome.

1:35:26
And Dale, you've shared lots of resources.

1:35:29
Is there anything else you'd like to shout out?

1:35:31
I think just to say that I feel in, in, in this thing, as with so many other challenges, I think we're stronger together.

1:35:38
We would just love people.

1:35:40
If you're not already part of an advice network, a wider advice network, please join Advice UK.

1:35:49
Please, you know, connect with other networks.

1:35:52
I, I think that that is a great way of facing some of these changes and challenges together.

1:36:00
So yeah, do join Advice UK, do connect with some of the great work that different organisations that that have been mentioned today are doing.

1:36:11
And Balji, anything you'd like to share before we wrap up?

1:36:15
No, just it was a great conversation.

1:36:17
Thank you for inviting me.

1:36:19
And I would echo what what was said by Dell, join your local partnerships, join local networks.

1:36:24
There's yeah, we can tackle it together.

1:36:27
Great.

1:36:28
Well, thank you.

1:36:28
A massive thank you to you 3.

1:36:30
This has been a really brilliant conversation.

1:36:31
And of course, thank you to all of you in the audience.

1:36:34
This.

1:36:35
I'm sorry we didn't get to all of your questions, but as I said, we're really keen to continue this conversation.

1:36:40
We're planning at CPAG to do future sessions around AI, including training.

1:36:45
So please do get in contact with us at [email protected].

1:36:51
Let us know if there's any topics you thought we didn't cover or you'd like to dig a bit deeper.

1:36:56
Let us know and we can design training for you.

1:37:00
Lastly as well, just to say we have a feedback form.

1:37:02
It'd be really helpful if you could fill this out for us.

1:37:05
It lets us know who's in the audience and again, just helps us to tailor our training as well.

1:37:09
Also just to mention, we've got our conferences in London and Manchester happening in October.

1:37:14
So again, we'll hopefully be having a few more conversations like this.

1:37:17
So if you fancy meeting up in person, please do take a look at our website for full details.

1:37:22
I'll always just thank us again for joining.

1:37:24
And thank you to DAO Lindsay Aboji for a brilliant conversation.

1:37:30
Thank you.

1:37:31
Thanks.