AI in its current state will never be able to progress beyond your own imagination

"What’s stopping me from building an automated NFT making machine?"

Probably nothing, as long as you have access to the API. I've also seen some interesting open-source implementations on Github.

I don't necessarily agree that "AI in its current state will never be able to progress beyond how you yourself imagine something to be." The whole point of DALL-E is for it to create something that can potentially EXTEND your imagination.

hey @mjsmmry , thanks for continuing this on.

I think the questions can get rather nuanced or even be paradoxical.

When we say 'can AI evolve past our imaginations?'. I just wonder if there are two angles to that question;
1.
Based on that which is imaginable can AI achieve something unimaginable?
It sort of poses the question, can you imagine that which is unimaginable?
Maybe the simple answer is no. Whatever exists, if it exists, or the thought of it exists than that is therefore imaginable. (But not to say that it would have been imagined, but it could have been).

Perhaps another way to formulate the question is;

Can AI solve problems in ways that no one could imagine? The short answer is, AI can ask questions that we ourselves haven't thought of. In that respect it can find answers that we may never have otherwise found.

We know that AI works by looking for repeatable and identifiable patterns using deep data analysis/neural networks. Say for example I'm an environmental scientist. I'm studying a river. The river fish are not healthy and the water has bad PH levels. I feed in a variety of data covering the last 10 years and I have water testing results and fish testing results from that same time period. Say for example the variables i feed in are those above but also charts of; weather (storms, rainfall, temperatures), no. of ships travelling through, the average sound of birdlife, dredging activities, industry runoffs, creek and storm water runoffs. The variables could also be further complex and detailed.

The AI may build an (almost) infinite number of possiblities searching for patterns that are so complex it would be almost impossible for a human to find the a correlation. Not that correlation is causation but it poses the question: "does this cause that?" The AI could propose a complex hypothesis based on repeatable data that humans could follow up with more testing.
Like it might find that 1 month after a storm occured and if dredging was done within 1 week of the nearest storm and industry run-offs occurred within 2 weeks of the storm starting that the PH rose to a level that affected the health of the fish, as shown by the spots reported on the fish, and the average sounds of birdlife declined indicating less healthy fish to eat and more unhealthy water.

So AI generates a repeatable identifiable pattern, posited as a question that humans could narrow down for further identification.

If that micro example of the fish is extrapolated; let's say AI gets super smart with all elements of the physical universe (physics; electricity, magnets, gravity, aerodynamics). And it can interpret physics experiments that have been created to the point that it can now run its own physics experiments (software-based) and learn what is useful, and grow on each iteration of those experiments. While it also has garnered knowledge from all over the textbooks, all off the lectures and the entire patent database plus other experiments people have run in software. What is to say that what it could conceive of, would not bend the rules of our wildest imaginations, of what is physically possible?

Scary? Wonderful?

What about if does the same thing with the drug industry? Could it cure all diseases?

What intellectual problems could it not solve?

If it were to do those things successfully, what about going beyond the physical, into philosophical, and religious questions. Could it give an answer that could convert/convince many people to a religious ideology?

Thank you, you are right. But if you combine people's imaginations to your own then at that point, it's a community. Doesn't belong to anyone. It thinks for itself.

Will designers and creative professionals become redundant?

interesting question I've been thinking about too. Most design designs, I've noticed, are based on the designs of already established design principles. Why are designers needed, when you could have an AI review the trends and produce a similar architecture.

Will AI evolve to produce full length feature films?

If they do, they will still need to meet the standards of the humans that watch them. If AI can beat Leonardo DiCaprio at acting or produce films on par with legendary producers, then I'm all for it.

Nice! I agree overall with your point of view...although I highly doubt a script would ever contain the exact same content as a book...but, sure, if I read the script, I'd probably imagine something different to how it visually plays out. I just find it hard to believe AI will ever really get to that point...where it's literally replacing human actors and directors? Call me a laggard, but I'm not sure I like the thought...