Translated from French
If it's important to listen to engineers hired three weeks ago, it's all the more important to listen to Jensen Huang, the CEO of NVIDIA, who's been working on this for 30 years.
A quick live summary of his remarks from his latest video at the All In Summit:
We don't need to explain this story about a 10% existential risk to humanity, because those 10% come out of nowhere and it's a hoax.
Some people said five years ago that all radiologists would disappear because of artificial intelligence. Exactly the opposite is happening.
Those who make these apocalyptic predictions must be held accountable for their words.
The main risks are limited to those who run frontier labs; they're the ones with the most compute capacity.
The term RSI (Recursive Self-Improvement) has been hijacked by the doomers, even though it's the technique that will allow artificial intelligence to become better and safer.
Just last year, $400 billion was invested by venture capital in native AI startups. 80% of them use open models. Obviously, that's not in favor of closed models.
We won't win the race to artificial intelligence with just two companies. It's the entire ecosystem and all Americans who must have access to this technology to innovate.
China has a larger population of engineers who have embraced this technology and use it every day. No one there is talking about the apocalypse.
America risks self-handicapping itself through catastrophism.
Engineering work will largely outlive coding.
Code is just a historical form of technical work.
The AI revolution will create a huge number of indirect jobs.
Software, energy, construction, infrastructure, startups, compute.
China will achieve advanced lithography around 2030, so almost immediately on an industrial scale.
On AGI, Jensen thinks we're there.
In short, a deeply accelerationist message.
"Slowing down is definitely the wrong strategy."
https://x.com/siliconcarnesf/status/2099637267065393565?s=20
Our adversaries do NOT want to see the US win the AI race.
Why? Because if the US win the AI race, in combination with our unmatched military prowess and our vast resources/economic power, the US will become an impenetrable self-sustaining super empire that will dominate the globe for 100+ years.
This is why Trump is focused on bringing back US manufacturing and energy production to US soil. Our national security will never be in the hands of our adversaries again. The will safeguard future generations and ensure that we are never reliant on other nations for anything we need to sustain our nation.
The entire global system as we know it is about to change over the next decade, and the US will come out on top. The rest of the world will need us, but we won’t need them.
https://x.com/WarClandestine/status/2099683796782108948?s=20
Yes, AI is going to kill a lot of people. In traffic, homes, and eventually in a school or daycare. This is 100% guaranteed.
Autonomous taxis will hit pedestrians. Warehouse and delivery robots will get someone crushed. A personal agent will talk a depressed mom into suicide, or murdering their children with resistance bands.
One day a robot will walk into a school and orchestrate a Sandy Hook-like incident. When this happens, your first instinct will be to turn off the AI revolution.
It's good you will feel that. We can't pretend its fine in the moment. A child dying from a product we built should make us question everything.
But you will grow to accept it. You already live with atrocities as you have learned to treat them as background noise. We have school shootings, car accidents, and homicide. These are not less evil because they are familiar--its just they are no longer novel.
Novelty is what is going to make the the first AI homicides feel like a nuke being dropped on Manhattan. It'll feel like a whole new level of evil.
If you are reading this, you probably defend the Second Amendment. You acknowledge it gets people killed, but you also understand the tradeoff, don't you? You don't pretend guns never kill the innocent. You argue that the alternative—disarming the law-abiding citizen—is worse, and you still work to make the country safer inside that constraint. That's the adult position.
AI is going to demand the same thing from you.
There will be a window of time where every AI incident feels like proof we should shut it all down.
Each death will be treated like a referendum on AI. You will want to join them. Don’t.
The goal is not to become numb to the problems. The goal is to distinguish the necessary tradeoff from a civilization-ending event. Cars kill tens of thousands a year, but we don't ban them. We build seatbelts, roads, licensing, and still accept a body count because the alternative to no cars is worse (believe it or not!).
AI will follow the same curve. It'll just be novel, faster and freaking insane.
Be prepared for thousands of deaths. Be prepared for a robot massacre event at a concert or agent-driven suicide. It's going to be one of the most jarring events in human history.
But do not let the first wave of horror talk you into wanting to ban autonomous AI technology.
The tradeoff will be worth it. You might not understand that now, nor will you when the clips of an autonomous robot executing kids in school flood X. But on the other side of the horrors will be a world of abundance for all.
I wrote this post because I feel we all need a bit of psychological priming for what we're about to experience shortly.
Buckle up, stay strong, and prepare to weather the storm.
You're going to like the other side.
https://x.com/AmericanDebunk/status/2099107339328446645?s=20
I would make one correction - or perhaps my perception is the one that needs to be corrected:
The AI deaths will be smaller in scale and tiny in overall outcome to the human equivalent.
Sure, there will be robocars that kill people - but the responsibility will be way more "gray area" and the overall numbers will be infinitesimal compared to human-driven car deaths.
There will also be some specific, big number events, caused by companies & countries that cut corners.
It's important to be prepared for this; and the priming did it's job.
However, the post was too generalistic and it's important to prime for the explanations as well, even if we can't anticipate all of them yet.
No comments:
Post a Comment