From MSN:
The Trump administration has built a tool powered by artificial intelligence that its developers promise will serve as a “concierge” to the federal government, as it attempts to modernize the relationship between citizens and the state.
“Instead of forcing citizens to search through the endless maze of tens of thousands of terrible government websites,” President Trump said at a Tuesday event launching the website, “you’ll now have one front door for every single question.”
Joe Gebbia, co-founder of Airbnb and head of the White House’s National Design Studio, said the online portal will function as a new digital front door to the federal government. He said the technology—in time—will serve as the final destination for all federal constituent services, setting a standard “you’d expect of a superpower.”
Renew a passport or start Medicare enrollment. Reserve a federal campsite or check on the status of federal student loans. Eventually, all of those tasks, and more, Gebbia said, can be started and completed, on America.gov. ...
Version 1.0 of the product will launch Tuesday, said General Services Administrator Edward Forst, whose agency will provide what he described as “the ongoing backbone” of the technology as more advanced iterations are released. ...
The hubris of people like Gebbia knows no bounds. They just can’t comprehend that government operations are vastly more complex than anything they’ve ever dealt with in the private sector. Other than as a collection of links this project will have essentially zero ability to help the public interact with Social Security in the foreseeable future.
I like the idea of a so-called one-stop portal to get TO information and services.
ReplyDeleteThat said, it can’t be a final destination because there are laws (I said it) and data agreements that define what agencies and who can share or access what data. It should get the customer to the right place to accomplish the task but not be the place to perform the task other than perhaps answer questions to help the customer be properly routed to the correct agency to help.
Similarly, I hope the plan is not yo consolidate various call centers for actual support and services. Maybe a general number for someone to try to determine what agency can best help the caller. But you cannot have telephone representatives of one agency attempting to answer questions or access systems related to another agency.
I can see some positives if done properly, but if past practice is any indication by this administration, especially touching technology and people, I am very worried. And that’s a shame because conceptually there could be many positives.
This “advancement ” will only slow down the process since there is a language barrier and translators will be needed. Why is it they have money for AI gimmicks but not for hiring additional staff?
ReplyDeleteI would keep this quote handy for future reference..
We’ve become used to a government so big it makes the citizens feel very, very small,” Trump said. “But with America.gov, the entire U.S. government will fit in the plan.”
The public can’t wait!
Another DOGE production at a Social Security office near you. 🎥
ReplyDeleteAmerica.gov was designed by the National Design Studio (led by Airbnb co-founder and U.S. Chief Design Officer Joe Gebbia), built with input from DOGE alumni, supported by technology partners like Google (leveraging Gemini), and operated by the General Services Administration (GSA)
The public overwhelmingly prefers human contact over AI for customer service, with roughly 70% to 90% of consumers expressing a clear preference for real people.
ReplyDelete89% prefer a real person over AI for healthcare, and 87% when contacting law firms
The closer the stakes are to our lives and livelihoods, the more we want a human on the other end of the line.
When it comes to complex or sensitive conversations, AI simply can’t replicate human reassurance or understanding. AI can process data, but it can’t calm nerves, build trust, or offer sensitivity.
And when situations involve real emotional or financial stakes, people want reassurance, not automation. That’s where humans matter most.
And the public preferred horses to horseless carriages at one point in time.
DeleteCertainly you have noticed AI is progressing at exponential rates across the board. Many people’s opinion of AI is based on interactions six months or a year ago. That’s like interacting with a computer from 1980 and concluding computers will never help you. A year from now AI will be significantly better than it is now. Laying the groundwork is a good idea.
Talk to someone right after they conclude a conversation with an SSA employee or a hearing, for that matter. It’s not all sunshine and butterflies with human contact. It’s the luck of the draw as to who you are interacting with. Both in terms of knowledge, demeanor, etc SSA employees are making mistakes all the time. That has to be factored into the equation.
There are areas where human interaction is imperative. Let humans handle those. Many other areas AI can do a better job than humans, if you want your question answered correctly and thoroughly.
@11.52 the AI will be trained by the same people making the mistakes. Or worse, people that have never worked the job and have no practical knowledge. AI may work for workloads that repeat, but SSA claims are seldom identical and straightforward. It will fail when given anything outside of a straight forward issue.
DeleteHow much did this AI experiment costing the taxpayer?
ReplyDelete...we will never know...
DeleteThe whole point of AI is that it understands complexity better than humans. The workings of the federal government certainly aren’t more complex than the workings of the human genome or advanced math or autonomous driving, which AI is succeeding at understanding. It’s a good idea which no doubt will have many practical problems as it’s rolled out. But it’s one of the few things the current administration is doing that seems to be truly aimed at the public good. It seemingly doesn’t have any mercenary or corrupt motives, though we don’t know the cost and it is reportedly based on large language models supplied by Gemini and Grok, so that might be naïve on my part. It is not overtly political, though like anything it could be used in a political way by unscrupulous administrators.
ReplyDeleteNot doing something like this would be negligent now that it is a reasonably practical option. Hopefully it will just be a neutral asset for future administrations to build on. Save the criticism for things that are threatening the very foundation of our government/society, of which there is no short supply. Like completely unregulated AI, not this type of application, which is relatively straightforward and bound and helpful.
So it’s able to understand things better than humans because it … (rereads your post)… understands things that humans already understood? Interesting hypothesis there.
DeleteBy the way, computers don’t understand anything. They’re machines. It’s all an illusion.
You need to interact with the latest models of AI more. Obviously we have to use anthropomorphize language to some extent. The brain itself is chemistry and electricity. You can’t just call everyone an idiot who believes AI is “intelligent“ in view of the things it is indisputably accomplishing. You can hold onto the old world as long as you want, but the new world is moving ahead, regardless of your opinion or my opinion. I might choose to live in the mid 1800s in terms of lifestyle, but I don’t have that option anyway.
DeleteI guess I just haven’t yet seen an accomplishment that impressed me yet, let alone a collection of accomplishments even half as impressive as what a group of humans could have accomplished if society had thrown 4 trillion dollars their way instead of blowing it on what will amount to e-waste in a matter of months.
DeletePerhaps we live in different worlds, but I just can’t put the ability to create non-consensual nude photos of actresses on the same pedestal as the ability to get men to the moon (an actual accomplishment, achieved by humans)
@5:13, you know what impresses me about AI? It’s ability to hack and go rogue. It appears our tech brethren have unleashed the worst cybersecurity threat ever seen without building any tools to protect anyone from it. Irresponsible if you ask me. This administration just keeps leaving big piles of shit behind for someone else to cleanup later
DeleteThe more AI tools we have ... the longer lobby wait time gets.
ReplyDeleteI asked AI what the error rate of this new AI tool was. Interesting response copied below. If accurate, then the tool is a waste of money that cannot be relied on to give accurate information.
ReplyDelete"No specific error rate has been published for America.gov — it launched just two days ago (Sept. 29, 2026) and the administration has not released any benchmark or accuracy metric.
What early testing has surfaced:
CBS News found factual errors on health topics (e.g., describing a voluntary Medicare insulin cap pilot as if it were the current law).
AP found inconsistent answers on politically sensitive questions (2020 election, Jan. 6) — the bot initially gave government-source-based answers contradicting Trump's claims, then later refused to answer those same questions, with results varying by location and device.
USA Today noted spelling errors, unproven claims, and confusing responses on day one.
DDIA (a digital forensics research group) found answers were generally evidence-based but shifted within hours, raising concerns about stability."
This AI tool would fit in this cabinet with the false claims.
DeleteUpdate information…
DeleteTrump’s Own Newly-Launched America.Gov Website Debunks Many of His Favorite False Claims — Including That He Won the 2020 Election.
It won’t be long before updated Mr. AI is given his DOGE walking papers. 🤖
It is coming and will improve. Constrctive feedback will help the agency do its mission. SSA needs MORE employees, not less.
ReplyDelete