From MeriTalk:
The Social Security Administration (SSA) believes agentic artificial intelligence (AI) is ready for mission use, but agencies will only realize its potential if they pair the technology with rigorous governance, continuous testing, and modernized data infrastructure, the agency’s chief AI officer said on June 26.
Speaking at ACT-IAC’s AI Forum, Brian Peltier, SSA’s head of technology, customer data, and AI and chief AI officer, said the agency is approaching a fundamental shift in how it delivers services as AI capabilities mature alongside years of technology modernization.
“We’re on the cusp of really changing the way we operate, how we work, and how we service our customers,” Peltier said.
Agentic AI is ready to support agency missions not only because the technology has advanced, but also because agencies have spent years modernizing their underlying systems, Peltier said. Even so, he cautioned that successful deployment depends on governance, security, and operational safeguards. …
24 comments:
Brian's saying what he has to, knowing that any public dissent from AI evangelism will result in him being out of a job and replaced by someone far less knowledgeable and responsible about government systems governance.
Where are all the haters now? Mr. Bisignano has the FrankAI show up and running STRONG. PAT is groundbreaking and can answer any question. IMAGEN turns hours of work into minutes. And ASC fills in the blanks where all else falls short. I am doing the work of ten men thanks to these tools, and we no longer feel understaffed. Now I’m just wondering if all the haters are all of a sudden feeling too sick to comment?
Too bad that DPS and other key software systems go down periodically throughout the day.
But we are ready for AI!
This “new” AI tool will be fully ready to implement in two weeks. I am firmly in the skeptical camp since many “innovations” have been disasters.
Later on in this same article, it talks about an internal AI tool for searching or querying policy guidance, and says that "frontline employees now use the tool daily to help serve customers." Is that true? I thought I had seen comments here that the tool was a joke and unusable... any frontline employees here who can share their take on the current version of this tool?
PAT is newer than the Agency Support Companion and is a lot better. It searches PolicyNet for answers I think, and ASC does not. But…it can still get things wrong.
Thank you, I needed a hearty laugh this morning.
@9:48. I tried searching PAT for a question I knew the answer to, to see what it would produce. It said no policy existed on that topic. When I knew for a fact it did.
Yeah, surely AI is a fad, certain to pass within a few years, even if the rest of the world adopts it. No use trying to utilize it to help with the massive workload everybody’s always complaining about. Let’s just do things the government‘s way, which has worked so wonderfully in the past. And one more thing – we want telework!
"I am doing the work of ten men thanks to these tools, and we no longer feel understaffed." And the work of how many women? I have followed this blog for many years and have never once heard a real SSA employee say they did not feel understaffed.
Well the rules are very complex, its hard for a human to process all the possibilities. AI would be trained by tech folks and they dont know the system and the policy. Also, for transparency who is checking that the info is correct and not biased against claimants? SSA currently struggles to do some basic stuff, anyone that has had a decision come in and go into pay can attest that many of those payments are incorrect, not by a few dollars, but thousands. Call to the FO yesterday, where I was told by SSA that a Claimant CANNOT have a representative for anything other than a disability claim. Making the workers dumber, relying on unproven, lowest bidder contractors to create an AI sounds like a disaster of epic proportions to me.
This gives Commissioner Frankenstein the go-ahead he needs to fire more SSA employees!
They are jumping ship at a fast rate already!
SSA says it is testing its AI case uses before and after deployment. Why does SSA not share those test results with the public? Presumably SSA knows from testing how often each AI case use hallucinates, misdirects people, or otherwise gives the wrong answer. Hopefully SSA is also running cost-benefit analyses. Reassure the public that SSA is being transparent by publishing that data, so a productive debate can occur about the best uses for AI at SSA.
IMAGEN is nothing but a character recognition and word search tool. It’s less an AI tool than AskJeeves or Dogpile were 30 years ago.
@10:20: Has it occurred to you that the reason your government hasn’t worked so well for a while because it’s chronically underfunded and understaffed? And is plowing ahead with tools that hallucinate (and cost more than adequate staffing would!) might not be a real solution?
Haven’t you read the Republican talking point that this administration is the most transparent in US history? You are violating the loyalty test that this administration demands from its federal employees.
@ 1:39 Yes, it has occurred to me. I’m sure I have more experience in government than you do, unfortunately for me. But I’ve seen how it works with my own eyes. The bureaucracy stifles creativity and innovation. Good luck trying to change the most obvious inefficiencies or inanities, even if they would obviously help serve people better. It has been next to impossible. The big changes that, against all odds, are accomplished typically take years/decades to implement.
To your points, AI will lessen the impact of understaffing. And if you think the presence of hallucinations in some contexts, which can be guarded against, outweighs all the obvious advantages of AI, keep doing things your way. Good luck competing.
Every industry, even the government, will be reckoning with the same choice in the next few years - do we consciously refuse to adopt AI tools that are inarguably better and faster in order to save jobs? Society as a whole has to come up with a solution to that unbelievably challenging and potentially destabilizing problem. SSA is going to get swept up in the same wave as everything else.
It makes no sense to focus on old school methods or carp about trivial issues when AI offers the potential to revolutionize government service.
And no, this isn’t Leland or Elon or Frank.
“Mother of all workflows” worked for me for 20 years.
I agree w 1:36. IMAGEN is not based on a large language model like what is available now. Not surprising since they started working on IMAGEN 10 years ago. It should be shelved the minute a real AI model is leased and tailored to SSA’s needs. It is more reasonable to judge what type of AI functionality SSA will have based on what a $20 prescription to ChatGPT provides, not IMAGEN’s clunky and limited functionality.
Try entering your SSA questions directly into ChatGPT (stripped of PII, obviously). It’s not perfect but it is surprisingly sophisticated and knowledgeable. It can also go beyond citing resources and provide explanations and suggest analytical frameworks. If it doesn’t make sense you don’t have to follow it, definitely double check its citations and explanations, but I think you will be surprised. And it’s only going to get better, since the frontier labs are pouring billions of dollars into it and they are in an arms race to capture market share.
Are they using the SSA AI version?
Tripadvisor’s AI Is Telling Vacationers That Gruesome Horror Grottos Are Wonderful Holiday Retreats
Planning your next getaway? You’d be wise not to trust the advice of an AI chatbot — even if they’re explicitly designed to draw on the testimonies of real travelers.
As The Guardian reports (https://www.theguardian.com/business/2026/jul/02/ai-summaries-tripadvisor-hotel-reviews-downplay-serious-complaints), an AI tool from Tripadvisor designed to summarize the reviews of hotels and other locations on its platforms routinely glosses over some of the horrifying complaints that guests have made about their stays.
The AI described one hotel in Cape Verde as “spotless,” failing to mention that it was being sued for mass food poisoning, according to the findings of the consumer campaign organization Which reported in the newspaper.
They are using the agency support companion (asc) to draft decisions and using policy assistance tool (pat) to answer any SSA questions
I will say this as a t16 cs being forced onto the phones for the past year... I was using Google to answer basic questions for people that i cant remeber off the top of my head. I have been using PAT now for a while and it's surprisingly good. Way easier then searching poms!
Post a Comment