a rock surpassed his intelligence shortly after his Shitter takeover, so, there is that.

Chess for example. There are many other use cases where dedicated machine systems are mentally superior to humans. Musk probably meant a more generalized intelligence- the current holy grail of artificial general intelligence, and as usual he is full of shit.

Yes, it can perform specific tasks faster than humans, and even better. But it has yet to make a discovery of which I am (or it is) aware. It just puts out something that has to be checked and tested by humans. (See: Poisonous AI recipes). Basically, AI is a sort of electronic idiot savant (roughly as that term was meant decades ago).

We will probably never see AI superior to human intelligence. Not that it can't be done, maybe, but because AGW will put an end to technological civilization before that can be achieved. Perhaps the next go-round/species will get it done.

human output also has to be tested and checked by humans (and machines). We make a lot of errors.

Humans do make lots of errors. But the humans beat AI for ingenuity and inspiration 100% to 0%. It's difficult to see those percentages changing any time soon.

is much quicker and less expensive.

Again, this is not general intelligence, it is ‘domain specific’.

it is more brute force computation. Similarly, a simple calculator can outperform a human easily in the speed and accuracy of arithmetic operations.

But yes, Musk is here again with his often BS prognostications.

Just like the 'turing test' got obsoleted as soon as even more 'brutal' forces arrived in LLMs. Dismissing intelligent systems for merely being 'brute force' assumes a definition of intelligence that is likely not supported by evidence.

have any reasonable characteristics of intelligence.

BTW, given their sometimes egregious faults, I doubt that any existing LLMs could pass an even mildly rigorous Turing test.

a superior game of chess. As I said, this was once a metric for machine intelligence. At chess a machine is literally smarter than humans.

Perhaps you meant consciousness not intelligence?

a lot is in the definition one chooses. Merriam-Webster sez:

1 a (1) : the ability to learn or understand or to deal with new or trying situations : REASON
also : the skilled use of reason
(2) : the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria (such as tests)

b: mental acuteness : SHREWDNESS

c: Christian Science : the basic eternal quality of divine Mind

2 a : INFORMATION, NEWS

b : information concerning an enemy or possible enemy or an area
also : an agency engaged in obtaining such information

3 : the act of understanding : COMPREHENSION

4 : the ability to perform computer functions

5 a : intelligent minds or mind
cosmic intelligence
b : an intelligent entity

IMO the best fit for defining "intelligence" are points 1 a & b and 3 i.e. reason, abstract thinking, shrewdness and comprehension.

Point 4 I guess would apply more to things like computer chess and LLMs or "AI" as it now stands in general.

Not sure what point 5 is about.

Is exactly what for example chess computers do. They are superior to humans because they are able to reason to many more 'plies' about the game situation than humans can. Dismissing this as 'brute force' dismisses all chess players at more than a beginner level, as the game, once basic skills are mastered, is about reasoning through the cascade of possible outcomes from any specific move.

I also agree with the first version of (1) " the ability to learn or understand or to deal with new or trying situation" - this is the AI goal currently known as 'Artificial General Intelligence' and the difference between this and 'specific intelligence', the skilled use of reason in a specific domain, is where I started from in this thread. AGI is indeed the distinct difference between what humans intelligence can do and what machine intelligence can do, right now, that we know of.

Perhaps we are in agreement?

Not just here, but in the larger debate. Every time a machine meets a given target for intelligence, society scrambles and says “ok, but that’s not *real* intelligence.”

Maybe the final test will be passed when a machine wrestles with anxiety, shame, and self-doubt.

And indeed it is likely to be civilization altering, on the level of what the printing press, electronic communication (starting with the telegraph), photography (and more importantly cinema and then television), and social media have and are doing to culture. The fact that it is being developed and deployed by giant tech monopolies, that it will be used to further consolidate wealth into the planetary oligarch class, is unsettling to me as well.

There are many types of "intelligence" and even within each category, humans may be experts in one area and idiots in others.

So then we make machines that can mimic or surpass human capabilities in one area but fail embarrassingly in others. How do we exactly define what's a "human-like" error rate? More to the point, computers make different kinds of errors than humans typically make. But the errors are what make us human.

And is there any point in making machines that mimick human errors?

Our pattern matching ability frequently hallucinates patterns that aren't real, but it has evolved this way because it is incredibly useful for us and other animals to adapt to new facts in our environment. It is perhaps a useful model for how an artificial general intelligence would also operate.

Given that there are indeed many types of intelligence, it doesn’t seem reasonable to declare that a machine isn’t *really* intelligent unless it demonstrates all of them. And although we probably shouldn't design machines to mimic human error, but it seems that at some point we’ll be saying “Well, they aren’t *really* intelligent because they don’t screw up like we do.”

The problem with all these predictions (AI joke, there.), is that current AI models are (at best) glorified auto-correct tools. They predict "answers" based on the data they've been trained with.

There's no intuitive leap, no spark of invention. Instead, everything gets mashed together into a slurry. Small wonder the results are weird at times.

Perhaps they're good enough to create mindless entertainment, but they're not yet good enough to replace human ingenuity and creativity.

That's my take, anyway. (Not that you asked.)

At this point, AI applications largely seem to be super-fast in decision-making involving an amassing of an ever growing and evolving database of consequences which are applied in a developing probability engine.

There is a less than zero chance that this will lead to true self ideation and contemplative awareness and when this occurs, watch out!

"Elon thinks AI will surpass human intelligence next year."

But nothing will duplicate nor surpass human emotions tho.

In every business venture that he's been a part of, he was either kicked out entirely (PayPal) or the business just used his name to government subsidies and higher stock prices. He's done next to nothing himself.

He's Elizabeth Holmes with a penis.

n/t

But all we have right now are faulty programs. Give up the ketamine Elon.

looking for more money for his xAI company.

next year too (as he has been saying for several years now).

By spreading misinformation.

So, yeah, fuck that guy.

It's easily surpassed most of the magat types.

If it doesn’t have the proper dataset for something, it can’t even make a recommendation. As said above, it is great at brute force computation, writing bits of code, fluffing up resumes, etc. But AI still can’t think on its own.

We are nowhere near the point where it could be considered a cognizant “being” of sorts. It doesn’t write songs or make pictures based on emotion/experience. It is not driven and it is not creative. It’s still dumb. Right now, it’s just a tool, and an excellent plagiarizer.

Watching these corporate "visionaries" warn about AI is a joke. Musk has been on of the loudest voices, but only because he's trying to raise questions about Microsoft's AI so his can have some time to catch up.

With Musk, it's always about making the most money an d having the most power. He already took over the majority of communications satellites in orbit. He bought and controls for his own whims Twitter. To those ends he is constantly out there critizing the competition and pretending he's being some futurist in the process when all he's doing is selling his own products.

Yeah, once Musk's own AI is the rage, he won't sut up about how amazing AI is for humanity.

Elon is an idiot.

I wrote my college papers on a typewriter with no spell check

microsoft word passed my by long ago