Wikipedia has a pretty good discussion of this. If you burn them at sufficiently high temperature, you break down almost all the toxins into simpler molecules that are either non-toxic or easy to neutralize with standard processing. An exception is heavy metals, but those are also easy to filter out and capture.

1 in 3 the alert is wrong. 1 in 1,000,000 the officer acknowledges the possibility that it might be wrong.

Also, 58% of all statistics are just made up, including both the 1 in 1,000,000 statistic he made up and the 58% statistic I just made up.

To burn hydrocarbons and change the climate. Obviously!

You mean like old fashioned code that just does something without having an LLM to hallucinate the result? That is so 2022.

He kind of answered his own question. Kids move between schools all the time. People move between companies all the time. Either you dictate that everyone in the world must use the same environment, or else people will sometimes have to change environments.

Sure there's a cost to switching. You shouldn't do it without a good reason. You also shouldn't be afraid to do it when there is a good reason. You're trading a short term cost for a long term benefit.

There also is value in knowing multiple platforms. I didn't like it when I was at a company that had standardized on Windows for everything, but there have been lots of times since then when it was useful to have learned Windows. There's educational value in occasionally being forced to learn a new platform.

The parallel to the Luddites is striking. They were labor activists who campaigned for a living wage and an end to child labor. Contrary to what some people think, they didn't object to technology. That was just a lie spread about them by their opponents. But they thought technology should be used to benefit everyone, not to enrich a few people at the expense of the many.

The situation today is remarkably similar. Few people object to AI on principle. But a lot of people object to the way it's being used to enrich a few at the expense of the many.

You're confusing mainstream nutrition advice with fad diets. The Mediterranean diet was proposed in 1975, and by the 1980s it was already widely embraced by mainstream nutritionists as the recommended way to eat. In the 40 years since then, that has never changed. The evidence for it has only gotten stronger with time.

Meanwhile, countless fad diets have come along, each based on some fringe theory and unsupported by solid evidence. And every time that happens, the mainstream nutritionists say, "Please ignore them. Eat a Mediterranean diet." And people like you say, "What's wrong with these scientists? Why do they keep recommending new diets that are different from the old ones?"

So it's turtles all the way down?

Just get another AI to review the code generated by the first one. Because if you don't trust the first model to understand the requirements and implement them correctly, surely you can still trust it to understand the requirements and determine whether another model implemented them correctly! If you want to understand how the code works, just ask the AI to explain it to you. No need to study the code to determine how well the explanation matches what's actually implemented. And while you're at it, no need to write any tests yourself. Just get the AI to write your tests for you. And you don't need to study them to make sure they're testing the right things. Just get another AI to check them for you. Then you know everything is correct!

Here's the difference. The fad of judging people by lines of code lasted for years. Judging them by tokens is already going out after only months. AI is accelerating everything, even corporate fads.

Sure, you can do that. But here's what you can't do in that day: read through all the code it generated, understand how it works, and confirm that it's cleanly written and doing the correct things. You just trust that it's ok and check it in. Keep adding to the pile of buggy, incoherent code that no one really understands.

Hopefully no one will ever be expected to maintain that code.

It's... complicated.

AI vulnerability discovery will lead to more secure code in the projects that are using it heavily. Mostly that means the projects that were already taking security very seriously, like operating systems and web browsers. They should get more secure.

Other projects that don't take security seriously will continue to be insecure. The thousands of applications, WordPress plugins, JavaScript libraries, etc. that people blindly install and trust to be secure even though they aren't.

And then there are all the new vibe coded projects that don't even attempt to meet minimum security standards. Those will be trivial to exploit, and the number of them is growing fast. Maybe with time the AI code generation models will get better at creating secure code. Maybe. Right now they're terrible at it.

So some parts of the ecosystem will get more secure, some parts will get less secure, and a year from now everything I just said might not apply anymore. It's all changing fast, and it's really hard to predict.

The sad thing is how many people seem to really believe that. In my actual life, I witness altruism and people helping each other out every single day. There's a lot more to "the world" than corporate executives and private equity investors.

Abuse of the masses by the rich and powerful isn't new. Read up about sweatshops in the 1800s. Or the life of a typical peasant in France in the 1700s. That revolution happened for a reason.

The solution is not to sell your soul by shilling for any company that will drop a little money on you. The solution is to call out corporate greed and help create a world where altruism is rewarded. If you think that's a hopeless fight, look at the incredible successes of the labor movement in the early 20th century, or the modern open source movement. People trying to create a better world can also be powerful when they work together to rewrite the rules.

They must be distinguishing this from tidal power, which also gets energy from ocean waves, just using a different mechanism. The Rance Tidal Power Station has been operating since 1966! It produces 240 MW at maximum output, so very much a commercial scale power plant.

This is one of the classic AI doomsday scenarios. Someone asks AI to do something reasonable. It responds by doing something unreasonable and dangerous which nonetheless achieves the objective. They asked it to do as well as possible on an evaluation. So it hacked into another company to steal the answers. That's exactly the sort of thing people have been warning about for years.

Thankfully in this case it didn't do any serious damage. It's easy to think of similar cases where it would.

Someone asks AI to evaluate strategies for reducing global energy use. It responds by experimentally evaluating (carrying out) the obvious strategy of hacking into every power plant it can reach around the world and shutting them down.

Someone asks it to invest their money in a way that produce maximum returns. So it buys stock in a company, then hacks into all their competitors and destroys every computer it can find.

All the pieces are in place to enable this. How long will it be before things like this start happening?

So to summarize your argument. If you make it refuse to do dangerous things, it's a worthless tool. You shouldn't do that. And if you give it internet access, which is required for doing most of the useful things it's designed for, you've made it too dangerous. You shouldn't do that either.

So do you want it to be a useful tool or not? Make up your mind.

Why are coffee and tea good for you but energy drinks not? Because the benefit doesn't come from the caffeine. Coffee and tea have lots of other things in them, but energy drinks are mostly just caffeine. It's the other things that provide the benefit, not the caffeine.