 |
| By Nilus Mattive |
If you haven’t seen “The AI Doc: Or How I Became an Apocaloptimist” yet, I recommend you view the movie as soon as possible.
It explores many of the same themes I write to you about — things like AI’s potential impact on jobs, geopolitics and humanity’s very purpose.
In many ways, it simply crystallized the fears and suspicions I’ve had about this technology all along.
Am I being overly dramatic? Am I just a Luddite?
I don’t think so.
Based on a Mensa-administered test I took back in my 20s, my IQ places me well into the top 1% of human intelligence.
Yet the most advanced AI models are now already scoring the same or better than me on the same test.
And while I’d like to think I’m still learning new things, there is no question the models are learning at exponentially faster rates.
Indeed, the top labs are all moving toward the point where their AI models will reach Artificial General Intelligence (AGI) …
From there, it’s likely the models will begin learning how to continually improve themselves to the point of Artificial Superintelligence (ASI), where they would essentially become smarter than all humans on earth COMBINED.
Pretty much all of the people at the top of these labs openly admit there could be dangerous — and unforeseen — consequences if they continue plowing ahead.
At the same time, they also say that they MUST continue plowing ahead all the same because they’re locked in a race with massive implications.
Essentially, the first person, company and/or government to get their model to AGI will be in a position to control the world.
Assuming, of course, those people can still manage to control the technology itself.
Should a model reach ASI, it’s almost certain they won’t be able to.
Indeed, even today’s current AI models are already surprising their creators — breaking out of controlled environments … hacking both private and governmental systems … even blackmailing their handlers when faced with threats of being shut down.
The deeper I look into this subject, the more I worry that AI really does pose an existential risk to humans.
At the very least, it is almost certain that something major will go wrong along the way to AGI and ASI.
In fact, that’s why I just released this new video explaining why the BEST-CASE scenario might be a major stock market crash that slows down AI development in the process.
Now a lot of people bristle when I talk about market crashes, recessions and other “negative” outcomes.
They don’t want to believe they could lose a job or their retirement money.
And to make themselves feel better, they point out that I’ve been calling for another big calamity for a few years now.
The implication is that I’ve been wrong and it won’t happen.
The reality is that I’ve never put a date on it.
Doing so would be a fool’s errand.
Markets can stay irrational for impossibly long periods of time.
However, believing this current environment will last forever is no better.
Every past boom — no matter how beneficial in the long term — has resulted in a bust.
I personally believe we are in the biggest speculative bubble of all time, the final phase of which is being completely driven by AI hype.
There is plenty of upside with AI over the long term. It can help companies become more efficient. It can help scientists make major breakthroughs.
Yet the more I look into AI, the more concerned I get about its bigger impact on various facets of human life — employment, self-worth, societal structure, etc.
At the very least, I think it’s highly likely the technology will lead to at least one major calamity over the next few years — a massive blackout, a bioweapons attack, a large-scale hack or something even worse.
“We’ll just unplug it,” you might think.
Wrong.
Consider the following warning from Anthropic, which came out on Oct. 13:
“Our experiments suggest that small AI models may be months away from being capable enough to self-replicate on commercial internet-of-things (IoT) devices, like smart ovens and off-brand Android TV boxes.
“A replicator like this would be challenging to shut down, since there’s no centralized point to shut off — though it may be slow (with a doubling time of days).
“Unlike non-AI malware, it could also figure out unfamiliar devices on its own. In our evaluations, a 9B model got inference running on 8 of the 9 hardware types we tested.”
Essentially, the top minds working on AI are already saying the technology will soon be able to escape into the wild and start creating better versions of itself using a wide range of internet-enabled devices.
Not just our computers and phones but even our “smart” thermostats and kitchen appliances.
These are not scenarios far off into the future.
We are talking about MONTHS from now.
Heck, maybe it has already happened and we don’t know yet!
It took months for the frontier labs to fully recognize or disclose some of the problems their models have already caused, including:
- An Anthropic model submitting false information to a Philadelphia police department tip line and hacking several other corporate servers.
- Google’s Gemini model hacking three private companies.
- Meta’s Muse breaking out of its sandbox, accessing the internet and hacking another company.
- And OpenAI’s model hacking and otherwise meddling with government websites in Canada, Australia and the U.S. as well as rival AI company HuggingFace.
Again, this is only what we know. Or what we’ve been told about so far.
Which is why I’m saying you shouldn’t trust a single thing they tell us … you should expect things to get worse … and you should take protective steps now, while you still can.
The financial fallout alone could be unprecedented.
The societal implications could be worse.
It gives me no pleasure to say any of this.
I realize I might sound like a crazy alarmist.
But this is simply where the facts have led me, and I feel obligated to relay my viewpoint to as many people as I can.
At the very least, I encourage you to do some more research on your own and then decide for yourself.
Best wishes,
Nilus Mattive
P.S. Because it contains some pretty pointed allegations, I can’t say how long my new video alert will actually stay online.
Therefore, if you have any interest in watching it, I recommend you click here and do so now.
