Hacker News Reader

Comment by arisAlexis | Hacker News Reader

The singularity. I don't think most authors of articles like these understand what the AI build up is about, they think it's another fad tool.

View on HN

You might want to listen to what people besides Sam Altman and Ray Kurweil say, at least once in a while.

You don't need to listen to those guys. The basic idea is quite simple. If human brains are basically biological computers, as electronic computers steadily improve and some point they'll surpass the biological ones. That point roughly is the singularity or whatever you want to call it.

It occured to me as a teenager and I wrote an essay on if for my uni admissions exam back in 1981 but it's not rocket science and the idea goes back at least to John von Neumann who came up with the 'singularity' term in the 50s.

That AI progress will stop because X? It's their normalcy bias (look it up) talking. There is no stopping.

Thank you for letting me know about “normalcy bias.” This is a term and concept I previously hadn’t the slightest inkling of. Now I understand the Light of our Machine God.

We've had AI booms before. In terms of capabilities, this one is following exactly the same trajectory. Human researchers come up with some sort of breakthrough improvement to AI methods, that results in exponential like growth of capability as we both solve the low hanging fruit the method offers at it, and scale up the compute and data available to the limits of what is useful for the method. Then, capabilities start to plateau, and there is a long tail of the new techniques being applied in specific situations as they get combined with domain specific tuning and architecture.

We are well into this process already. Core chat capabilities have pretty much stalled out. But most of the attempts at application are still very thin layers over chat bots.

IMO, a big reason for this stagnation is that many researchers who could push the AI further choose not to because they came to believe that AI will do no good for humanity. It would be odd to have intellect strong enough to advance AI and at the same time somehow to not see the consequences of such 'progress'.

Do you have any proof of progress plateau and any proof that we had been in the past in an exponential AI cycle or you just think that it is so?

It is totally irrational: let's invest in AI companies, on the off chance that Singularity happens and makes money and entire economics system obsolete.

There is no winning scenario.

It's not an off chance. Technology and innovation doesn't stop, only with nukes.

Indeed, the critics can only be so critical because they are not convinced of the revealed truth that we are materializing machine god. How irrational.

I'm hoping that it's sarcasm to be invoking a machine god while calling the non-believers irrational.

While the majority of people invoking the idea of a machine god are being irrational the idea in itself has at least some merit. Some animals are more intelligent than others. Humans far more intelligent than the vast majority of animals, and with tools and cooperation in large societies we live in an existence they are not equipped to comprehend. This invites the question of what are the limits of intelligence. Is it far beyond our capabilities? If so, what does that look like to us.

Even if you believe in the Singularity, there will be a crash if investors stop investing before it hits.

Why would they stop investing if they think correctly that AI progress is unstoppable

Reply on news.ycombinator.com