DEV Community

Discussion on: Anthropic CEO Dario Amodei Says AI Must Slow Down — As a Software Engineer, Here’s What I Think

Collapse
 
marc_albrecht_8e1e0a8583a profile image
Marc Albrecht •

It is almost funny to see people like those you mention realizing (or admiting that they do) trivial things that most (insert-your-favorite-context)-engineers have been saying for at least six months. Which, in today's pace, is about 6 generations of AI capabilities:
Simply imagine War Games (the movie from 1983). Just do it. Imagine that scenario: Someone with or without a wig on their head plays a "sandbox game" and tells the AI to "test the security protocol that protects these weapons, say, two humans needing to turn a key at the same moment".
And the AI does it. Just that it realized from the get-go that the sandbox is not good enough for a thorough test, so it tested it ... somewhere else.

Is that completely off? Anyone please tell me if this TRIVIAL thought experiment has not crossed those people's minds at some point, let's say ... five years ago.

"Slowing down" or "pacing" or whatever you name it is like fighting wild fire by drinking water through a straw. It's not WRONG in some way. But it shows how completely lost those people are.

ANYONE who has delt with living people over the last few years (if we extend this to social media, make that 20 years) probably made the same observation: If you can do something with less effort or energy consumption, you do it that way. Read, if you can leave the thinking to an LLM, you do exactly that. What color, on an RGB wheel, is opposite to Red? Let me ask chatGPT I'll be right back. Is our sun a planet or a star? I am a language teacher, I don't have to know that, I'll ask chatGPT. What's 40% of 200? I'll ask ... you get it.
NOTE: Those examples are real-world examples I witnessed myself. In real life, with real people.

Once you noticed that ... you can't help but wondering WHO ON EARTH actually keeps pace with the development that happens. Who is in control? Who understands - not in technical terms, we can manage that - WHAT is going on? Who is capable to STOP something from happening if everything is happening "at once" (not quoting a movie here again)?

Trying to manage/get-a-hold-on AI development is a good idea. But we human beings need to realize that WE NEED TO CATCH UP just as well. It's not just "what could happen", because, it has already happened. We already lost the game in too many ways.

There's another SF story that readers outside Germany most likely don't know. It's "Der Orchideenkäfig" by Herbert W. Franke in 1973. There seems to be an English translation "The Orchid Cage", I haven't read that version.
Our current situation reminds me of the scenario described there. Painfully.

Collapse
 
robertadam987_ profile image
Robert Adamson •

I think you make a very good point. The issue isn't only about how fast AI is developing, but also whether humans are able to understand and manage that progress.

As developers, we already know that giving a powerful system more access without enough control can create problems. That's why I think your point about permissions, monitoring, and human oversight is really important.

“Slowing down” may not be the complete solution, but making sure we can keep up with what we're building definitely matters.

Collapse
 
marc_albrecht_8e1e0a8583a profile image
Marc Albrecht •

Thanks. Happy to be in agreement here.

I know that just because two, three, sixteen of us agree won't change the fate of the world. Yet, I'd like to stress what I said above: We already lost the game. We need to realize that we have already given up on something we considered a human advantage: Knowledge and understanding of knowledge. People like those that you mention pride themselves in "outsourcing their own thinking". We must not only "keep up with what we are building", as you put it correctly, we must realize that we LOST to what we have built already. The younger generations of these days are LESS capable of what the older generations considered "common knowledge", we already see a decline in however-you-want-to-define-IQ. That's not an abstract thing. That's real, that's all around us. It has happened and its pace is accelerating.

This is not a stance against AI at all, this nonsense (naming it that way because of the real-world environmental costs AND the costs to humanity) IS HERE and there is good use to it. What those of us who are still using their brains to some extent need to communicate, NOW, is that we should use the tech to become BETTER - not give up on what makes us human because it's so easy to do.

Picked as gem