PUBLIC TESTIMONY
Testimony by Kevin De Liban, Founder of TechTonic Justice
3:07:30
·
3 min
Kevin De Liban, founder of TechTonic Justice, provides a cautionary testimony about the risks associated with implementing large-scale integrated technology systems for public benefits. He draws on his experience as a legal aid attorney and warns about the potential for significant failures and human suffering.
- Emphasizes the high-risk nature of overhauling eligibility and enrollment systems, citing failures in various states
- Recommends careful problem definition, scoping, and evaluation of technology capabilities before proceeding
- Suggests implementing strong oversight measures, including public reporting, pre-deployment testing, and phased implementation
Kevin De Liban
3:07:30
Good afternoon, Sharon.
3:07:31
With the evidence and the other members of the committee.
3:07:33
My name is Kevin De Leon.
3:07:35
I'm the founder of an organization called tectonic justice to fight the ground level harms, their artificial intelligence algorithms, and related technologies caused low income people.
3:07:45
I come to this work after 12 plus years as a legal aid attorney in Arkansas whose specialty was public benefits involving in at points the integration of these various eligibility systems.
3:07:56
The short version is beware, run the other way.
3:08:00
This is a warning, the sky is falling, and all of the other good things.
3:08:04
Similar projects, oftentimes headed by, you know, large private vendors in various other states, have resulted in enormous failures.
3:08:14
And these states vary.
3:08:15
It's not just Arkansas, Rhode Island, it's Colorado, it's Texas, places of different sizes, different political climates, and everything else.
3:08:22
And when they fail, the result is outstanding and unsupportable human suffering.
3:08:29
Right?
3:08:29
People who are unable to get benefits so that they can buy food or get health care or do many of the other things that they need to survive.
3:08:38
So these are incredibly high risk uses.
3:08:41
There hasn't been an instance yet where a jurisdiction has successfully somehow overhauled their eligibility and enrollment system to integrate multiple applications without it entailing significant risks and failures.
3:08:55
If this is the path that that New York City is gonna go down, I have some, I guess, cautionary guidance.
3:09:02
1 is making sure that the problem is defined and scoped in detail.
3:09:06
What is it that you're actually trying to solve for?
3:09:08
What problems?
3:09:09
Is this technology capable then of solving that problem in a way that is efficient, effective, fair, not harmful?
3:09:19
And cost effective.
3:09:23
If you can't commit or if you're gonna go this way, will you commit to the resources to minimize the harm and ensure proper oversight?
3:09:30
That means making sure agencies have the sophistication necessary to interrogate vendors, contract well, run all sorts of tests and projections, consulting community members that are gonna be affected in an ongoing way that actually enables them to have power and to participate meaningfully.
3:09:50
Importantly, it's necessary as part of this to have extensive public reporting and forcing the agency to answer questions that might not be answered in their own internal documents and the things that they choose to make publicly available.
3:10:04
Of course, freedom of information laws have to be strong around this so that the public can have access to the information that has to be predeployment testing.
3:10:12
There has to be phased implementation so that it's not applied to everyone at the same point.
3:10:16
You have to maintain the non AI the previous way of doing things so that you can turn that switch off and get people benefits the previous way.
3:10:25
That concludes, I think, by 3 minutes, happy to offer any additional information or answer any questions.
3:10:30
Thank you for your time.
