I am guessing you mean AGI? They usually mean they’re not generally intelligent, i.e., displaying cognitive flexibility for different types of tasks without training or extensive fine-tuning. Think of a human intern, you don’t need to tell that human intern beyond a simple phrase when you need something. The intern will figure out how to do it, including figuring out what they don’t know, and what they need to learn to do that thing.
Sadly, a lot of AI products are going to be bad and overpriced. But the good news is that since there’s less investment floating around, stinkers won’t stay for long.
Quality is a subjective property unless you’ve established protocols, frameworks or standards which allow objective measurements.
Software engineering never had any strict standards or requirements (beyond borrowing abstractions from computer science) unlike other types of engineering. You never needed any license to be a software engineer, which showed up the in the “quality” of candidates or employees. The closest you get to ensuring even quality is education or training, but even that is a wash because of the varying quality (hah) of programs. Many programs also fudge up rankings, so you have shit-tier schools (and faculty) pretending they’re top-20 or whatever, meanwhile they are global zeros.
Now it doesn’t matter at all, you can’t put into place standards and requirements you never had to begin with just because your employment is on the line. People have been shipping broken, terrible shit for years because parts of it worked, or you got customer lock in for whatever reason. Microsoft Teams is a great example of this.
Software engineering, and computer science by extension, don’t require any formal degree or licensure anymore specifically because of AI. Thank God for that, actually, because a lot of terrible-at-their-field people got ahead because of sheer luck and we are stuck with them as though they’re the “experts”. In that sense, we should celebrate AI as an equalizer. It’s a digital gun, people should learn to use it, and responsibly.
> The email is dated 26 Jun 2016, right after Brexit, and in it, Epstein writes:
> return to tribalism . counter to globalization. amazing new alliances. you and I both agreed zero interest rates were too high, as i said in your office. finding things on their way to collapse , was much easier than finding the next bargain
> This is a perfect example of what Naomi Klein calls "disaster capitalism."
Disaster capitalism only works if there’s enough law and order to preserve capital for those who acquire or control it. People who believe in ideas like the two mentioned in the email have experienced nothing but post-WW2 America-enforced global security, albeit relative amounts of it depending on how you are.
That security is not a constant guarantee, it is not a natural element of the world, it is not a force of physics that exists by itself. In a way, the post-WW2 order was crafted by centuries of both physical and mental toil. Undoing it will not bring the selective prosperity or success these people hope it will.
If you decide to be “enlightened”, you force your enemies to assume the same posture. If you decide otherwise, then what they do in return is fair game. Then very soon afterwards you find out there’s no such thing as sheltering from the aftermath of such events. Look at any societal collapse example in history, recent or otherwise.
This is why the ultra rich have been building out heavily fortified, private, self sustaining compounds. They know the potential consequences of their choices, and are fine with them as long as they can leverage technology to insulate themselves from them. It's basically Fallout mindset.
Societal collapse doesn’t necessarily mean all-out nuclear catastrophe, and can also mean transitioning from one set of norms or rules establishing the status quo to another. People shouldn’t labor under the false impression that they will have guaranteed success when such transitions happen. What are they modeling this on, exactly? Countries where mafias are de facto rulers, or there’s some level of dictatorship, are only held up due to external pressures put on by democratic world orders. As mentioned, guns and bombs can guarantee protection up to a point, and the rest of that necessary work is done by ideas and ideals.
Secondly, I am afraid a lot of the people who believe they’ll have some sort of success in scenarios with lesser law and order than they grew up with will find out it wasn’t their winning personalities which guaranteed them anything. There will always be someone more brutal ready to just take yours and dispose of you. To put it more plainly, if all they like is your money then you’re just standing in the way of that.
I would say don’t do what this article is saying. Have the agents write to a space dedicated for their edits specifically, keep the agents.md file as your instructions to them. Not your task-specific prompt, but any generalizable actions across different tasks.
reply