TL;DR - if it can't provide consistent rersults nor be held responsible, "AI" will never replace humans.
E
Long version:
LLMs regurgitate previously spewed stuff (including from other LLMs) which they pretend to analyze.
Unfortunately this analysis yields inconsistent responses, so an LLM's answer to YOU may be different
than its answer to me. Further, sometimes they just create fiction our of nothing, which we RATIONALIZE
and PRETEND is "hallucination."
Software algorithms do not hallucinate. Software algorithms provide the same output given the same input
each and every time. 2 + 2 ALWAYS EQUALS 4. Any algorithm that is not consistent with this concept has
the flaw of nonrepeatability, nonreproceabliity, is nonscientific, and can't be trusted.
And then there's what thing about holding people responsible and accountable. Example: If you're the director
of human resources for a company, then your policies about how to deal with employees that don't show up should
be consistent. If they are not, well that's the nonrepeatability (above) but also you get to be held accountable. So
if you let Alice not show up on Mondays but you tell Bob he'll be fired for not showing up on Mondays, and tell Charlie
he has to work 80 hours a week... well at some point (lawsuits notwithstanding) you'll have to answer for your behavior.
AI is not held accountable. It can't and doesn't feel pain. You don't "fire" AI leaving it wondering where the next paycheck
will come from to pay for its home, spouse, children, vehicle, etc. You just retart it and a new "instance" is there to screw
your company again.
These are also reasons why Tesla will never ever achieve full self driving (FSD). Insurance companies and law enforcement
both want one neck to choke, and they have the most expensive lobbyists after pharmaceuticals. Insurance companies won't
insure a car that has no human driver they can punish (higher fees, non-renewal, cancellation, responsibility for accidents, etc.)
and law enforcement officers (LEOs) won't allow a driverless vehicle to drive on public roadways because they have nobody to
cite, ticket, file charges against, or remove privileges (driving, parking, etc.)
AI is just the current moniker for "broken software that is inconsistent, provides semi-random results, is GASLIGHTED as hallucnations,
and is neither scientific nor repeatable. It is aritficial in that it is man-made. So is poop. Both stink. It is not intelligence, as it lacks
various factors like self-evaluation, self-improvement, and any internalization of anything it is asked to do, does, say it did, or should do.
There are various factors in play, and while I mentioned lobbyists for insurance companies and LEOs, there are also "lobbyists" for the
"AI" industry. Those are called fundraisers, and just like CryptoBros, QuantumDudes, BlockchainBuds, and all the other fads in the past
decades, they exist to part fools from their money. Then those fools become evangelists. That doesn't make Crypto, Quantum, Blockchain,
Bitcoin, or any other other fads any less real. They exist to take money from people who call themselves "investors" but really they're gamblers.
The difference is an investment is a real interest in real things (property, money, rights, etc.) whereas gambling is buying something today
for X dollars hoping to sell it later for more than X dollars. You own nothing, have nothing, but you do have the right to swindle a future sucker.
Sam Altman is a smart guy - he gets people to give him money. That's not "genius". It's PT Barnum. Mark Zuckerberg may be a genius, but
his neurodiverse nature makes it difficult to accurately know. Jeff Bezos is the idiot who wanted to sell books online, and accidentally created
a backend that allowed international sales of various products by many many people. Fortunate for him (and his mom) but it wasn't the plan.
AI - not a thing. It's regurgitated spew in a mess on the wall and nothing is sticking.
But... one other thing is... companies that use AI to replace humans are just as stupid as companies that say other stupid things:
- Please listen carefully because our menu options have changed. Wait, WHEN did they change? Why should I have to hear this crap every time I call you since 1994?
- If this is an emergency please hang up and call 911. Sure, I'm calling the Doctor's office and I have an emergency but I'm too stupid to call 911 so to "reward" me I have to listen to this stupid disclaimer. Either connect me OR offer me "hit 9 to be connected to 911." (For non north American users, we use 911 as the short phone number for emergency services. You may know it as 999, 112, etc.)
- We are currently experiencing a high volume of calls. No, you're not. You are inadequately staffed. Stop lying and hire real people to talk to your customers.
If you've heard these messages and thought "oh well that's life" you tacitly allow these incompetent buffoons to keep doing that very same thing.
AI is not the problem. It's also not the solution. It's also not intelligent.
I hope you all have a great week. Monday is the start of it. AI doesn't exist. If you're still unconvinced, go tell ChatGPT to timestamp every answer and watch it lose its timekeeping marbles in less than one minute.
