One of the debates in AI is regarding the question as to whether AI will accelerate fast enough, potentially combined with brain-interfaces, that a super-elite class could form.
Ray Kurzweil, world’s foremost futurist, says no, and makes a good point about cell phone adoption rates. I will broaden his idea. Every new technology essentially increases its rate of adoption globally. Cell phone adoption is now at 91% in 2023, which is incredible if you think how many decades it took for older technology to spread, like cars, telegraph, and even television. In fact, when I was in Belize once, other than a bare minimum $5000 home, the shirtless workers had only two things on them as we drove by: A machete and a cell phone. The jumps for new technology skipping old ones is real.
With ChatGPT being the fastest product adoption in history at 100M active monthly users in within two months of launch, we can see the rate of tech adoption increasing, so when advanced AI, and AGI are developed, won’t most of the world adopt them instantly? “For context, it took Instagram two and a half years to get to 100 million. TikTok got there in nine months.” – Time. Seems likely.
However, here are a few points to consider:
If the rate of technology growth continues, then the rate of increase in “computer intelligence” or human-like capabilities will increase faster than we realize, and here is the math.
Computation power of machine learning models has doubled every 5.7 month, which is much faster than Moore’s Law which doubles every 1.4 years.
So, if a machine has the equivalent IQ of say 100 today, then in a year, it could be 400, since (100*2)*2), far surpassing any mortal alive
It’s one thing to be capable of such high scores, but another to actually benefit from it. This depends not only on using it, but on having direct uninhibited access to it, which is probably not going to be the standard. Another component could easily be the limit of which one can interface with it via brain-interface tools. Realize that most people using ChatGPT are playing with it, some use it for work, and a handful of people are trying to leverage it into mega-capabilities right now.
In another year, the IQ may increase at the same rate (400*2)*2, or 1600, which if connected to the right tools, perhaps a circular loop, with ChatGTP and numerous plugins, start creating AGI.
What becomes challenging then is the rate at which these are adopted, and the first developers to employ such power.
If you still do not believe this is likely consider the following. On one hand, the world overall has become far more equal in wealth over the last century, however, in the US and other advanced economies most data suggest it’s actually decreasing. More importantly, you have the largest difference in wealth between the top 1% and the bottom, possible in recorded history. So, yes, the world is good and fair, but I am concerned about the risks of an AI super-elite, who could exist as individuals or governments. In fact, many of the world’s richest and most powerful people are probably not even known since they do not run publicly traded companies. Take for example the guy that trades up to 5% of the NASDAQ every day.
Yes, we will all become quite rich faster than ever, but the those at the top could accelerate beyond our reach, especially risky considering the top may be permanent unlike in history, combined with life extension tech that is just over teh horizon.
- I think a possible story of this AI acceleration would be this:
- Some kids develop the first self-learning AI and
- escape to live on a boat where they figure out how to make billions overnight using quant trading.
- They evade governments by using the AI to create the best AI-guided attorneys while living on boats, and
- figure out how to manipulate govt themselves,
- which govts are trying to take control of it for their own power-hungry purposes as well.
- The developers escape capture from military pursuits using the AI’s predictive abilities and engage in simultaneous global hacking of all military networks;
- then using AGI tech to hack the minds of people turning them into semi-hypnotized slaves, or at least socially engineered on a mass scale.
- The world turns off their computers to avoid the digital plague, and
- we revert to farming and communities for survival.
Not sure how the story ends, but something like this seems quite possible, if not an underestimation of potential fantastic (good or bad) outcomes.
Although the pro-tech voices out there clamoring for the coming intelligence explosion, I have said for years that I suspect we may see the opposite: a mass revolt or upheaval when advance AI/AGI arrives when people perceive it as a threat to their own existence; does not also the very survival of humans predict it in some ways? Sure, some think we will merge, but what about those that choose not to?
HumanIT 