91探花

In a recent and rather provocative statement that鈥檚 got the AI community pretty fired up, Sam Altman, CEO of OpenAI, made what might be his boldest claim to date 鈥� indeed, according to Altman, we鈥檝e 鈥渁lready gone past the event horizon鈥� when it comes to artificial intelligence (AI).

Whether intentional or not (I鈥檒l take a wild gander and assume the former), this is the kind of phrase that鈥檚 bound to attract attention. It sounds simultaneously exciting and ominous, but what does it mean?

Well, if you鈥檙e not familiar with the term 鈥渆vent horizon鈥�, don鈥檛 worry, you鈥檙e definitely not alone. It鈥檚 actually a phrase borrowed from astrophysics 鈥� an event horizon marks the point around a black hole where nothing, not even light, can escape. In other words, it鈥檚 a tipping point, the point of no return, so to speak.

Now, the idea of a 鈥渢ipping point鈥� or 鈥渢he point of no return鈥� is a little alarming no matter the context, and there鈥檚 no arguing the fact that having the head of one of the world鈥檚 mot influential AI companies suggest that we鈥檝e not only reached but actually passed this point with technology is pretty darn scary.

He didn鈥檛 say it outright, but Altman鈥檚 comment suggests that artificial intelligence has reached a stage of development so significant that its trajectory is no longer fully within our control. He doesn鈥檛 pose this as a big warning for humanity, striking fear in the hearts of people around the world (who may or may not have watched 鈥淚-Robot鈥� one too many times). Rather, he actually asserts that he personally thinks the outcome will be a听鈥済entle singularity鈥�. But still, it鈥檚 kind of concerning and definitely something to ponder.

So, given Sam Altman鈥檚 expertise, position and experience, what are we supposed to do with this information? Is Altman correct in these assertions? Or, is the OpenAI CEO making grandiose claims to add a little fuel to so-called 鈥淎I fire鈥�?

But, first things first, let鈥檚 democratise the complexity of this verbose tech talk and explain a few concepts so that we can all be equally terrified or calm, depending on our conclusion.

What Are AGI and ASI?

听

Alright, so AI, ordinary artificial intelligence as we know it 鈥� language models, image generators, recommendation algorithms and so on 鈥� are very different to what we鈥檙e referring to when the future and potential of AI becomes concerning. In those contexts, but people are actually talking about is AGI and, in extreme cases, ASI.

AGI is Artificial Generative Intelligence and ASI is Artificial Superintelligence. The former goes beyond narrow AI systems and is capable of learning, understanding and even reasoning across a broad range of tasks at a level that is equal to the ability of humans in the same contexts. It鈥檚 able to not only do things like write code or play chess, it would have the ability to actually learn a new language, invent a whole new game, debate philosophical concepts or even manage a business 鈥� it would be able to do all of this without needing to be retrained or receive specific data inputs.

Now, this is an incredible idea in itself, something that we haven鈥檛 yet achieved 鈥� unless you鈥檙e Sam Altman in which case you may think differently.

ASI, on the other hand, goes even further than AGI. At this point, the AI doesn鈥檛 just become as intelligent as humans, it actually becomes way smarter than humans in every way imaginable (as well as in ways we can鈥檛 even imagine). ASI would have the ability to solve problems we haven鈥檛 even thought of yet, optimise entire economies and innovate at a pace far beyond human capacity.

So, when we think of sci-fi movies in which robots are taking over the world, these are the two types of AI that are actually being referenced, most commonly ASI. Unsurprisingly, both concepts are intrinsically terrifying.

听

听

Why Altman鈥檚 Making Headlines Once Again

听

That鈥檚 why Altman鈥檚 assertions that we鈥檝e already crossed the threshold 鈥� already moved into a realm in which achieving AGI and even ASI is possible 鈥� have caused a fair bit of commotion. He鈥檚 not saying we鈥檝e reached AGI just yet, but he believes we鈥檙e heading in the right direction and we鈥檙e well on our way.

According to Altman, developments in machine learning and neural networks are happening faster than anyone predicted a decade ago. Tools like ChatGPT, voice synthesis, autonomous agents and multimodal systems have rapidly advanced our expectations of what AI can do. And, with every breakthrough, we鈥檙e inching closer to systems that resemble AGI in both behaviour and potential. While we鈥檙e not there yet, Altman鈥檚 comment suggests we may have entered a momentum phase, where progress builds upon itself, while, at the same time, human ability to halt or fully steer clear of it is quickly diminishing.

A pretty scary prospect, if you ask me.

So, why is Altman making these claims? Are they worth worrying about or is he simply making a strategic PR move to spark conversation? He is the CEO of one of the world鈥檚 most successful AI companies, after all 鈥� it鈥檚 his job to keep OpenAI relevant, and I鈥檇 argue that this is one hell of a way to do that. All publicity is good publicity! And all those other age-old mantras we hear so often in the world of marketing.

There鈥檚 no denying the fact that someone like Altman has pretty serious insight into AI tech, far beyond the ordinary person, as the leader of OpenAI 鈥� his professional thoughts and opinions are, by no means, irrelevant. However, having said that, many think that this particular statement is mostly about just stirring the pot, for lack of a better term.

If you consider the thoughts of other very experienced and knowledgeable experts in the field, the opinions are varied, with some believing that we鈥檙e still many decades away from AGI, others adamant that it鈥檒l be upon us in the early 2030s and another whole camp steadfast in the opinion that AGI, nevermind ASI, isn鈥檛 possible at all. So, if nobody else can agree on this, how is Altman so confident in his opinion? But more importantly, what actually听is听his opinion?

Well, the answer, some believe, is that he鈥檚 not really saying as much as we may think he is. Altman鈥檚 statement about us having 鈥渁lready passed the event horizon鈥� was intentionally as vague as it was ominous. It鈥檚 an assertion that鈥檚 bold enough to cause both experts and laypeople to stop and think, but it鈥檚 still blurry and imprecise enough for it to slip under the radar and allow him to avoid garnering the title of being 鈥渦nrealistic鈥� or 鈥渇ear-mongering鈥�. Is he really saying anything particularly novel?

Because, while his statement seems pretty terrifying and sci-fi-esque at first glance, what it really means, if we take it to its most extreme extent (for the sake of argument), is that we鈥檙e kind of, probably heading down what might, potentially be the path that could, one day in the future, maybe lead to AGI, or something vaguely similar to it. But, isn鈥檛 this something we鈥檙e already somewhat aware of?

As Dr. Lance Elliot, a world-renowned AI scientist, puts it in an article recently published by Forbes, Altman has consistently changed his definitions and timelines associated with AGI and ASI, making it difficult to follow his predictions for the future and somewhat tough to interpret the CEO鈥檚 approach to understanding what these prospects mean for humanity, both on the positive and the negative sides of the coin.

And, no matter your title, experience or expertise, who鈥檚 really to say what the future holds for humanity with regard to the development of AI or anything else, for that matter? Dr. Elliot references a pretty fitting quote by Franklin D. Roosevelt, who said, 鈥淭here are as many opinions as there are experts,鈥� and in this case, it鈥檚 probably wise to bear this sentiment in mind.

Ultimately, while the notion of humanity passing the 鈥渆vent horizon鈥� may have seemed like a big, bold statement at first, many are leaning more towards that fact that in reality, Sam Altman has managed to use a grandiose, poignant phrase and statement to say a whole lot of nothing we didn鈥檛 already know. Why? To get us talking.

And hey, hats off to Altman. After all, we鈥檙e talking, aren鈥檛 we?