On the moral merits of AI investments
Short/abstract: the (lack of) morals, when a company, an executive team, or even CEO, go to a financial institution, a hedge fund, a wealthy investor, and solicit $200B, $400B + to the intake of $1.5T (and counting!) over time (all the “big” tech/AI companies are on the hook for roughly this amount), and effectively having one technology, and promising another.
What they have: access to, and leasing of, supercomputers, which they use in conjunction with LLM's (large language models), which can do a fair bit of impressive computing – “brute force” style. Throw intensely powerful hardware at it, “fuel” it with knowledge pools and sifted/stolen data – watch “magic” happen.
What they promise (investors): Actual Magic™ in the form of software that will fix it's own code, write newer versions of itself, replace all human-reliance on the building/development of current and new software (and hardware). Come up with it's own hyper efficient methods of implementing and designing hardware and software to work quickly and cheaply. (all this the lie of GenAI)
It's like if I told someone I had a Ferrari in the garage – but not just that, a flying Ferrari, and once the cash was exchanged, opening the garage to show only a moderately decent sedan – not useless, but not too great, either.
Parables aside, tech co's know what they're doing. They know the hot phrase of GenAI will never materialize.
The phrase fraud comes to mind
Part of the AI Notes series. Previous entries here