Over the decades it’s been my experience that people new to the Social Security world dramatically underestimate the complexity and sensitivity of the work that the Social Security Administration does. I wonder how long it will take for DOGE employees to figure this out. I doubt that their arrogance can long coexist with knowledge of just how complex Social Security is.
As an example of the complexity let’s imagine a recent widow calling in to ask about benefits she might be able to receive. Sounds like that would be a common sort of transaction and it is. Here’s some of the questions that should come up and there are plenty more that may come up:
- How old are you?
- Do you have any minor children?
- Do you have any disabled adult children?
- Are you working and, if so, how much are you earning?
- How much income of any kind do you have now?
- How much do you have in the way of resources, such as money in the bank?
- Are you disabled?
Depending on the answers to those questions and potentially more, the widow and members of her family may be entitled to these sorts of benefits and she and others in her family may easily be entitled to two or three of these at the same time:
- Aged widows benefits
- Disabled widows benefits
- Child benefits
- Disabled adult child benefits
- Mothers benefits
- SSI
- Retirement benefits
- Disability Insurance Benefits
19 comments:
This exact scenario should be posed to Frank. He seems to think the vast majority of calls are akin to returning an item at Amazon.
Look who is acting all arrogant. Every one of us at SSA started at ground zero (well the DDS folks that came to SSA had a leg up). Experience takes time. I am not defending DOGE, but over decades being at the agency have seen politics and folks they bring in come in wide eyed. I take the tack in using these opportunities to share knowledge and leverage their clout to make improvements that have meet headwinds in the past. I do not always align with the past and present politics and approach but want to take advantage to move programs forward where I can.
That is why the high drop-out rate for new Benefit Authorizers and Claims Authorizers in class occurs. SSA is so complex and the various benefits so entwined, it takes years to learn the basics and become proficient. Like it has been said, "We are not at all like Amazon."
The highest respect to Charles for running this extremely valuable blog. But this post is just plain out of touch with what AI is currently able to accomplish. The complexity of Social Security policies is exactly why state of the art AI is needed for SSA. There are many SSA employees in field and hearing offices who just aren’t up to the task of accurately applying those regulations and offering customized advice and responses to individuals, especially under time pressure. Errors and oversights happen all the time, not to mention the delays involved. AI will make SSA more reliable, not less. I am referring to employees using AI as a digital assistant, not AI displacing employees. And the public should always have an option to handle its business the “old-fashioned” way. Somehow the conversation around AI on this blog seems to have been reduced to the current phone system’s flaws. That is a small fraction of how AI can help further SSA’s mission. Not to mention that AI is improving at exponential rates, and many flaws today will likely be gone in six months to a year. At least if SSA is able to get its hands on state-of-the-art capabilities.
Also, there is no reason to conflate AI and DOGE, nor reckless cost-cutting personnel actions with stepping back and thoughtfully using AI to provide more accurate and efficient service by our current workforce. There is a difference between the political versus the substantive actions that have been taken or are being contemplated. DOGE didn’t invent AI and to be honest it seems like most of what DOGE has been doing in terms of combing through databases could have been done by circa 2000 software. All the BS about dead people getting benefits has nothing to do with AI, it’s just running database queries and not being too bright or careful about how you interpret the results.
Many of the obstacles of the past, such as legacy systems holding back software innovation, can in fact be addressed to some degree by AI. AI is capable of understanding and producing better computer code than all but the most brilliant humans at this point, and can do so faster, by orders of magnitude. It can play a helpful role in modernizing entire codebases. No such tool has been available up till now.
I don’t like it anymore than you do, but the fact that all of humanity is going to have to accommodate and adapt to AI is obvious and is becoming more obvious by the month. Humans will still be in the loop, it will just be at a higher level, at least for the near to midterm future. They will be able to exercise their higher faculties (including interacting compassionately with the public) in their jobs more often, as opposed to being buried under mental grunt work. Over time fewer employees will be required but it seems to me that simply letting attrition run its course will be sufficient to “right-size” the ssa workforce for the next few years at least. Society as a whole is going to have to figure out what to do with all the “excess employees“ eventually.
For now it is best for SSA to concentrate on making the most effective and reliable and equitable use of AI, rather than pretending that this technological inflection point in history will spare us. I’m talking about long-term SSA employees with deep knowledge and expertise doing everything they can to help implement AI intelligently. I’m not talking about turning the whole shebang over to DOGE or any other tech component that doesn’t have knowledge of our policies and the practical issues employees face in the field or the effect of our actions/inactions on claimants and beneficiaries.
Ok Leland- calm down.
Can I have what you are smoking? It must be some really good stuff.
My genuine question: how likely is it that AI is implemented at the levels you speak of without a massive IT budget investment? The COSS says we’ll have it in a year, but when the budget is not increasing enough to cover inflation I just don’t understand how it can be done.
The Department of Government Efficiency has prodded federal employees to take buyouts, demanded they share the weekly “five things,” shut down numerous programs, and shaped mass layoffs across dozens of agencies.
Now DOGE is asking some feds: Care to join?
Multiple employees in recent weeks have received unsolicited invitations to interview with U.S. DOGE Service. The email typically comes from an unsigned “USDS Recruiting” .gov email address.
“I am reaching on behalf of the United States DOGE Service (USDS),” the email starts. “When applying for a federal government position, you noted that you were willing to share your resume with other federal agencies. After an initial review of your qualifications, we would like to invite you to an initial 20-minute phone interview to share information about USDS, the types of roles we’re looking to fill, and learn more about your background and related experiences.”
@2:23 it’s called information. You’re right, it’s good stuff.
@ 3::43 Good question and I don’t know re the costs. I do know that there is an arms race among massive corporations to get market share,, which should drive costs down. Ultimately it will depend on whether Mr. Big can deliver on his promises related to AI. It wouldn’t be the first time that SSA fails to deliver software solutions that private industry uses every day. But the potential positive impacts for SSA in this situation are qualitatively different than anything previously possible. It’s not just nibbling on the margins of what we do.
It's even more complex than you present, because many people do not even know why they are contacting SSA. The most basic example is the use of the term "SSI," which I would wager 85% of public misuse.
Anyone working with the public at SSA can interpret and understand the claimant means some type of payment/benefit they're receiving, but if you go purely by them using the term "SSI" more often than not, they're not on T16.
Secondly, people call with bizarre scenarios you cannot even decipher until you review their account and check recent notices they have received or payment history. People call constantly about Medicare supplements SSA does not administer.
People will say they didn't receive their monthly payment, and it's not even their payment date for another two weeks, etc. And this is before even considering language barriers and interpreters.
AI has a long to go. And as the original post talked, determining if someone potentially qualifies for a benefit is very inquiry intensive.
Here's the full story if anyone cares to read it: https://federalnewsnetwork.com/federal-report/2025/06/doge-sends-federal-employees-a-recruiting-pitch/
For those who say ai is the future, call Verizon with a technical issue and see that it will take you 15 minutes to speak with an agent. That is if you don't get hung up on or stuck in an infernal loop.
I’ve heard rumours Scott Coulter (the DOGE CIO) is out; any confirmation?
He is no longer on the org chart. I guess that is confirmation?
This is interesting about Scott. Can anyone confirm?
Say AI can only unambiguously decipher 25% of the calls coming in. That’s 25% fewer calls for the employees. As long as they include a feature that channels the call to a person when the caller chooses to do so it’s clearly a good idea.
Just add a real “speak to live employee” option. Problem solved. Any company or organization that doesn’t do that is incompetent. “Don’t be stupid” should be SSA’s motto. Many issues are solvable if simply directed addressed. Why they so often aren’t is a mystery.
Michael Russo is now listed as CIO again, but there is a rumor he will be replaced (again).
Post a Comment