I have been following thew development of AI for a long time. I am a nerd, I love technology and I have spent many hours boring the pants of my friends and colleagues about AI over the years and persuading them to watch Alpha Go (none of them ever have). I have used ChatGPT to write my SEO content since it came out in late 2022.
I was astonished back then and I couldn't believe how little interest my peers showed in the articles it had written for me from a simple prompt. I'm no sage, I'm orders of magnitude more stupid than the people at the cutting edge of this technology but even I could see how powerful this technology had just become.
Now, in the space of what is little over a year the capabilities of these AI models have grown exponentially. Already, in most areas of intelligence these machines are wiping the floor with all but the very smartest humans. As a species we are already hanging on by a thread albeit linearly in each of the disciplines.
This won't last. The goal is AGI of course (Artificial General Intelligence) also known as the singularity. Its a fuzzy term with no clear definition but however you measure it, in a nutshell it is human parity intelligence in all fields. I think like everyone else we thought this may be several years away but I just don't think that anymore. This discipline is accelerating at light speed.
We are not just walking towards AGI we are galloping towards it.
The thing is, no-one really knows how it works at a fundamental level. Its not called a black box transformer for nothing. But I won't bore you with the details of how it works, no-one reads these articles anyway. But its not good. Not by a long shot.
If you are interested you can read how it works here from this smart guy Kiel Dang, watch this snippet from the 60 minutes show with Geoffrey Hinton "Godfather of AI' here and then listen to this excellent Youtube episode from TRT with Connor Leahy
Then go over and read this article by George Takei of all people about Mark Tegmark, a super smart machine learning guy.
'Tegmark observed, “If you’d summarize the conventional past wisdom on how to avoid an intelligence explosion in a ‘Don’t-do-list’ for powerful AI” that it might start with these four don’ts:
1. Don’t teach it to code
2. Don’t connect it to the internet
3. Don’t give it a public API
4. Don’t start an arms race
Here’s the bad news.
On each of these dont’s, we’ve already done it.'
Listen to these guys and you realize that there are no brakes here, there is no safety net and there is no chance at all that we are going to pause and evaluate this technology. We are going to get to AGI sooner or later and when we get there, there is no way we will be able to control this thing at all. None. At every level it will manipulate us, out smart us and outplay us and it won't even be competitive. As the guys found with Alpha Go Zero it plays the game in 4D and we can only play in 2D. We won't even be able to see it coming.
The upshot is that we have just fertilized an egg, now we are incubating it and soon we are going to give birth to a God.
To put it crudely we have bolted into the womb of Pandora's box. Now we must wait until it opens to discover whether we have whether we have created God or the Devil.
Either way it seems to me that it will be calling the shots.
Comentarios