Are GPTs the way to AGI, probably not
In an opinion piece for the NY Times Gary Marcus indicates why he has reservations on the future of LLM GPT AI systems.
Silicon Valley Is Investing in the Wrong A.I.
“Buoyed by the initial progress of chatbots, many thought that A.G.I. was imminent.
But these systems have always been prone to hallucinations and errors. Those obstacles may be one reason generative A.I. hasn’t led to the skyrocketing profits and productivity that many in the tech industry predicted. A recent study run by M.I.T.’s NANDA initiative found that 95 percent of companies that did A.I. pilot studies found little or no return on their investment. A recent financial analysis projects an estimated shortfall of $800 billion in revenue for A.I. companies by the end of 2030.
If the strengths of A.I. are truly to be harnessed, the tech industry should stop focusing so heavily on these one-size-fits-all tools and instead concentrate on narrow, specialized A.I. tools engineered for particular problems. Because, frankly, they’re often more effective.”
Points I’ve also been making here several times over the past few months, along with others about the perilous state of the current US economy and how the “Current AI Hype Bubble” could be a disaster for it.
But the question of what is “Artificial General Intelligence”(AGI) is something that has at best had an elusive answer akin to “Shoulder shrug handwaving” and impossible “What ever you want it to be” type statements. It’s something that a group of 33 specialists from 28 institutions have got together to try and address more reasonably,
They come up with,
Definition : AGI is an AI that can match or exceed the cognitive versatility and proficiency of a well-educated adult.”
Which although it sounds profound is actually not that useful.
Because the use of,
“match … Well-educated adult.”
Is not actually a useful measure.
It’s been pointed out that the “use of aids” “dumbs us down” in that it causes us to “loose skills”. I first heard this when I was in school. With first electronic calculators and whilst still in school computers.
Whilst many would argue that it’s not important or even irrelevant, it is true that certain skills are not developed because of the use of aids.
What most do not realise is that those traditional skills that are seen as nolonger worth teaching due to the ubiquitous use of aids, are actually important. Not for what they directly teach, but indirectly teach. That is they give new viewpoints that are force-multiplier tools that enable us to reason in either new ways or to levels we otherwise might not.
At the end of the day the two things that have moved humans forwards over many thousands of years are,
- Stored Knowledge.
- Use knowledge to reason.
They were and still should be the foundations of becoming “Well-educated”.
Sadly as gets often observed these days, producing “Well-educated adults” appears to be nolonger a goal of the education system in a number of Western Nations.
