[zwkrt]

I might sound a little bit 'tinfoil hat' here, but I believe that what follows is not hyperbole. AI is already the 'Architect' more than most of us would like to admit. Even if it is not sentient, the various AIs that we use during the day were designed with a purpose and they are goal oriented. It is worth reading Daniel Dennet's thoughs about the intentional stance--we know that a toaster is not sentient, but it was designed with a purpose and we know when it is or is not achieving that purpose. That is why we might sometimes jokingly say that the toaster is 'angry with us' or that when the toaster dings that it is happy. It is actually easier for us humans to interact with objects when we know that they have a purpose, because that is similar to interacting with other humans who we know to have purposes.

Coming back around to AI, ChatGPT was designed with a purpose, and people project intent onto it. People act like it is an agent. And that is all that matters. The same is true of the Tiktok AI, the AI that calculates your credit score, the traffic lights by your house. Hell, it's also true of your stomach.

The point is that objects in our environment do not have to be literally conscious for us to treat them as conscious beings and for them to fundamentally shape the way that we live and that we interact with our environment. This is pretty much the basic tenet of cybernetics. To believe that all of these tools do not have intention and that they are 'just tools' used by some people to influence other people is not wrong, but I don't think that it captures the richness of the story.

Differentiating where humanity/consciousness begins and where the technology ends is already more complicated than most people think. Traffic lights train us just as much as we make traffic lights. I fully believe that people will be saying "this isn't true AI, it doesn't /really/ have feelings" long after the technology that we create is so deeply embedded into our sensory I/O that the argument will be moot.

[Bjartr]

That's part of why there's objection to claiming sentience, it distracts from the discussion of impact by dragging a whole lot of extra philosophical baggage into the conversation when it's not yet necessary.

[guhcampos]

That's kind of exactly my point when I say they are not architects, just tools: I agree 100% that people project intent into these things, and I believe that's exactly what our "ex Google employee" is doing here - and it's dangerous.

It's dangerous in part exactly because it shifts the responsibility for the acts of the tool to the tool itself, and away from its author. Like deforestation was the machine's fault, not the fault of the humans driving them.

I can never agree with your affirmation that "AI is already the Architect". It is not, the AI does not design anything. It does not plan anything. It has no ideas, no critical thinking, no judgement of value or morals. The AI just does what it's told, like a tool, a worker ant, or any other algorithm. It's complicated enough that it's not obvious to us what it was told to do, but ultimately it's all it can do.

[zwkrt]

I understand your point, and I'll agree to disagree. I think it just has to do with what we value. Even though it is a tool, tools change what options we have. If you have a shovel, 'digging holes' is now an optional activity for you to pursue that wouldn't have been otherwise. Is the shovel the 'architect' of your ability to dig holes? Maybe no, but the tool-human interaction is a back and forth. Tools generate affordances and humans choose to act on affordances.

Maybe put it this way, if there was an AI that could plan out your day in a way that would optimize some metric of happiness that you agree with, you might start to use the AI. Is the AI the architect of your day because it plans it out and tells you what to do, or are you still in charge because you could choose to stop using the tool, even though it would not be in your best interest?

I think this is the point that we are reaching with AI: it is a tool that is so flexible that it doesn't just offer single affordances, but begins to be used as a guiding function for what decisions to take. At that point I think it /is/ an architect of some kind.

Again though, this is mostly just quibbling about definitions and terms.

[la64710]

What about a system emerging from another system? https://clementneo.com/posts/2023/02/11/we-found-an-neuron

[imranq]

la li lu le lo