Caroline Selman

So we can't say what's happening because we don't have the data. But the risk of bias is roundabout. Like you say, certain ethnicities or other characteristics. So, for example, if you... have something that can be treated as a proxy for a particular nationality.

So we can't say what's happening because we don't have the data. But the risk of bias is roundabout. Like you say, certain ethnicities or other characteristics. So, for example, if you... have something that can be treated as a proxy for a particular nationality.

For example, the individuals might be being disproportionately picked up and identified as somebody to investigate for fraud, not necessarily because there's any more risk that that individual would have committed fraud, but because of how the algorithm has worked in terms of what it's using for proxies to pick people up and identify things.

For example, the individuals might be being disproportionately picked up and identified as somebody to investigate for fraud, not necessarily because there's any more risk that that individual would have committed fraud, but because of how the algorithm has worked in terms of what it's using for proxies to pick people up and identify things.

So the key concern is whether tools that are being rolled out might be disproportionately picking up individuals and investigating them for fraud when they haven't committed fraud.

So the key concern is whether tools that are being rolled out might be disproportionately picking up individuals and investigating them for fraud when they haven't committed fraud.

Yeah, so I think, first of all, with sanctions, the starting context that looks through all of this is, again, it's a system that doesn't work and causes harm. And what those figures show is that if you're black or mixed ethnicity, you are more likely to be impacted by that system which doesn't work and causes harm.

Yeah, so I think, first of all, with sanctions, the starting context that looks through all of this is, again, it's a system that doesn't work and causes harm. And what those figures show is that if you're black or mixed ethnicity, you are more likely to be impacted by that system which doesn't work and causes harm.

And there's still lots of gaps in terms of that data, but that raises really serious questions for DWP about who is being impacted, how and why. We have some statistics now for sanctions, but it's a concern across universal credit and how decisions are being made in relation to social security.

And there's still lots of gaps in terms of that data, but that raises really serious questions for DWP about who is being impacted, how and why. We have some statistics now for sanctions, but it's a concern across universal credit and how decisions are being made in relation to social security.

So again, like those considerations about if you're ruling out automation, that risk that you're potentially baking in some of those things that might be pre-existing bias and not having the ability to properly check for that or prevent that from happening.

So again, like those considerations about if you're ruling out automation, that risk that you're potentially baking in some of those things that might be pre-existing bias and not having the ability to properly check for that or prevent that from happening.

Yeah, so I think we've talked already about the real need to properly reform the sanctions and conditionality regime. But there is a fundamental point about some of these things needing quite a big culture shift and fundamental change. And that is,

Yeah, so I think we've talked already about the real need to properly reform the sanctions and conditionality regime. But there is a fundamental point about some of these things needing quite a big culture shift and fundamental change. And that is,

could include looking at, for example, in Scotland, they're trying to take a more rights-based approach, inclusive of kind of radical things like listening to people and making sure that the evidence is at the heart of things as well. But alongside that, there are things that could make things a little bit incrementally better now.

could include looking at, for example, in Scotland, they're trying to take a more rights-based approach, inclusive of kind of radical things like listening to people and making sure that the evidence is at the heart of things as well. But alongside that, there are things that could make things a little bit incrementally better now.

We would support there being much more transparency about things around about evidence. So some of what John has been talking to, I think what it demonstrates is the sheer amount of effort that he and others have had to put into and how difficult it is for individuals on the outside to be able to evidence and demonstrate that systems aren't working.

We would support there being much more transparency about things around about evidence. So some of what John has been talking to, I think what it demonstrates is the sheer amount of effort that he and others have had to put into and how difficult it is for individuals on the outside to be able to evidence and demonstrate that systems aren't working.

And it is quite difficult and intensive to do that from the outside. So as a starting point, a culture shift in terms of being open and transparent about, you know, when you've had an inquiry that it doesn't take several Freedom of Information Act requests and decision notices requiring you to put that stuff out in the public domain, that that is put out in the public domain.

And it is quite difficult and intensive to do that from the outside. So as a starting point, a culture shift in terms of being open and transparent about, you know, when you've had an inquiry that it doesn't take several Freedom of Information Act requests and decision notices requiring you to put that stuff out in the public domain, that that is put out in the public domain.