The headlong rush into artificial intelligence (AI) these days produces a steady stream of new players. As a concept though, AI dates back several decades, with a simple mission of using computer science and robust datasets to enable problem-solving.
The mission naturally evolved over time, leading to formation of Open AI in 2015 and eventually its 2022 launch of ChatGPT, the chatbot that seems to be the signal that AI had arrived!
What’s the goal of AI?
Put simply, AI is intended to efficiently solve problems, answer questions, and make decisions based on the analysis of large amounts of data. That certainly suits the world of Social Security, with the need to understand options amidst thousands of rules and tens of thousands of pages of arcane documentation explaining these rules and their myriad qualifiers and exceptions.
One would think the use of creating machines that can understand and communicate with humans in natural language would clearly demystify the complex and cumbersome world of Social Security and make it simple to apply the rules to everyday circumstances.
Maybe, maybe not.
A Real Life AI Failure
Our AMAC Foundation Social Security Advisory Service recently handled an inquiry from a caller asking for a second opinion. He had posed a relatively simple question to ChatGPT: If my wife claims her retirement benefit early, will she be able to switch to half of my benefit when she reaches full retirement age?
ChatGPT said sure, no problem, and that would have been correct if the question dealt with survivor benefits rather than retirement benefits. The point that was missed in the data crunching was that the question was based on the “file and suspend” loophole that had been removed years ago and replaced with the “deemed filing” rule stipulating that, when filing, the wife is claiming all benefits available to her, with reductions due to filing before full retirement age (FRA). This rule would preclude her from getting fully half of her spouse’s FRA benefit amount.
No detail is too small
The correct application of Social Security rules is a situational process, with nuances based on each individual’s circumstances. AI shows promise, of course, but we may still be quite far from the point of simply accepting machine-generated decisions without question. In the situation described here, the caller was wise to seek a qualified second opinion, and our Advisory Service is a logical place to provide that!
To learn more about the AMAC Foundation’s service, visit AmacFoundation.org/Services.
This article was unintentionally funny for multiple reasons. Not the least of which are the known issues with the accuracy of at least some of the datasets used in the LLM building process for multiple versions released to date. So, trying to rely on what ChatGPT may or may not spit out depending on how a specific question is posed would be iffy at best. So, depending exactly on how a query is carefully worded, in very specific terms, it can generate quite different or even completely inaccurate outputs. Testing ChatGPT to try and successfully navigate the morass that is SS or any other equally convoluted government program is something you will have to wait for the development of a general intelligence AI. Which is something much more powerful than say ChatGPT and about 3 yo 5 years away. Of course when we build and releae that broadly across the global Internet, the least of your concerns will be SS queries as AI will be running almost everything.
Two examples of internet/ mobile phone system errors right here in this comment — and both were done After the comment was posted. I always check what I wrote before I post it – by simply reading it — Twice — before I post it. In the first sentence the word – to – is missing between the words – lead and more- so it should read — ” errors may lead to more errors ” And ” the ” is printed twice in the second sentence – It should read ” It is good to know about the AMAC Foundation’s Social Security Advisory Service.”. There is a Principle involved in bringing these errors to the attention of whoever the comments are sent to – it is respect for Truth. This sort of practice by the internet / mobile phone system is corrupt and shows incompetence on their part – not at all in the interest of Truth .
Important article Gerry Hafer, artificial intelligence is no substitute for intelligent people and it is something that needs to be watched carefully — because of the possibility of it making errors and those errors may lead more errors — if people actually have trust in that technology to the point where they think AI is nearly infallible . It is very good to know about the the AMAC Foundation ‘s Social Security Advisory Services. . I am 74 and I am truly thankful to God for having developed a high level of knowledge, intelligence , skills and abilities many years prior to the internet.. Designing and making tools since the 1970’ s including parts for navigation instruments , weather instruments and microscopes and having respect for the rules of English Grammar as well as the standards for engineering practices – the whole computer, internet , mobile phone related technology is something that I do not need in many ways.. Regarding the changing of the English language in so many ways– to promote the internet way of thinking. — I don’t want any part of it. English Grammar is important and the rules should be respected – for many reasons pertaining to Truth, accuracy , basic survival for example – describing geographic aspects of an area, clearly, like the location of clean safe water sources. Many internet instructions are as twisted as a corkscrew — and the more serious a situation is the more need for clear, intelligent communication – made possible by the rules of English Grammar.. Knowledge of history will help to set things right – it will help to describe what defines Truth and Liberty as well as the Principles that hold the Nation together.. In the spirit of respect for Faith, Family,and Freedom. God bless America, land of the free, and the home of the brave.
,
:
AI, simple put is a farce and is very dangerous to humanity, even more so than anything else.How can humans program a pure AI with all are petty prejudices inherit to humanity.The old programming saying sums it up perfectly.Garbage in Garbage out.