Hacker News Reader

Comment by Trackmal | Hacker News Reader

I am stunned by these results and I find it absolutely amazing, but somehow this also makes me kind of sad. So many things we consider worth learning are loosing their value with these advances. It really makes me think if I should stop trying to improve my skills, because in maybe 20 years there is not much left what can not be done better by a machine. :/

View on HN

DALL-E has been trained in the styles of creative humans. It can mashup the content that it has been fed but I haven't seen it create a new style of artwork like impressionism or pop art that it hadn't already seen as examples. It also has no understanding of what is generated. We are impressed as human observers because it appears to understand the task we presented.

How many hours have you spent working with it so far? You obviously have a lot of experience with it, so I’m wondering how you drew this conclusion.

Zero. But I did read the research paper, web site and viewed the web site mentioned in this hacker news post along with the samples. I've worked with lots of machine learning models and tools and understand the underlying design of the system. You can't ask Dall-E to create a pop art picture of Corgis if you didn't train it on images of Corgis and pop art. I'm not downplaying the achievement of the system that creates an incredible connection between the input images and the text descriptions. But.. at the end of the day. It is not creative in the same way humans are creative.

That’s just begging the question. We don’t know if DALL-E is creative in ways isomorphic to humans. It might be. How would one go about structuring a hypothesis? Have you read the original DALL-E paper?

Worth considering: https://arxiv.org/abs/2112.04035

So what would make up a Turing test for creativity?

This is a fascinating question, I'd imagine many other people smarter than me have thought about it. Specifically in the domain of the visual arts.

I think it’s vitally important to distinguish between skills you do for fun and skills that earn you a living.

For job skills to earn a living, you should definitely be paying attention and adjusting in order to future proof. Learn skills that will not become obsolete.

For skills for fun, don’t bemoan the fact that an AI can do it better. There is still inherent value in you learning and enjoying how to paint. Not everyone needs to be the best in the world. It would be silly to say “I love baking cakes for my family, but there’s no point because of British Baking Show.”

> So many things we consider worth learning are loosing their value with these advances.

Should you not learn math because of the existence of calculators?

Compared to people who lived before the widespread adoption of calculators, you should probably limit how much time you spend getting good at calculations on paper and in your head, yes.

The point in life has never been to be "The best" (if it is, you've always almost certainly been bound to fail); it's to enjoy life while doing your best, enjoying the experience of life wisely. It actually makes me hopeful people will be driven to finally realize this when we're close to obsolescence in terms of abilities, and then we maybe can progress as a civilization toward a better society (using the newfound abilities).

I don't know--I'm super optimistic! Not only is it incredible in a way I haven't found technology incredible for a long time, but it has the potential to disrupt and simplify the hugely labor- and talent- intensive creative process.

Today, if you need a logo made or some clipart for your web page, to do it the correct, legal way, you have to either get lucky with some stock artwork or scout around for an artist, evaluate portfolios, select a few and buy some samples, decide on one and iterate back and forth until you have something you like. Then you have to have legal agreements in place, make sure rights and copyright and royalties and all that shit is decided, be careful about how you use that art (do I have the rights to put it on a billboard too?)

Imagine a far future where any creative work can just be freely generated with a text description, and the output is unencumbered by IP rights. Type something in and get an infinite scroll of outputs, select one, and you're done.

Extend it to all sorts of media: Music! "Two minute upbeat song about lawn care, jazz style." Out pops an infinite scroll of jingles. "Lullaby for 2 year olds about dogs." "20 minute opera in German, about cycling, in the style of Mozart." Movies! "Three part superhero series where the main character walks backwards." "Romantic comedy but with talking turtles." "Sci fi movie about underwater colonies with a shark villain."

This could be the future if intellectual property lawyers don't fuck it up with artificial scarcity and "digital rights" like they fucked up the copying of bits across the Internet.

The output of an ML model isn't clearly unencumbered by IP rights. There's already prompts to DALL-E that clearly output a copy of a Wikipedia image it's memorized, and the publicly available smaller models have lots of prompts that output images covered in literal Shutterstock watermarks.

This isn't real AI and it didn't come up with these images by imagining them. It's a blob of every image on Google Image Search stuck together in a way that's managed to differentiate between them (in the calculus sense).

Scroll through an infinite amount and be done? Crazy.

Drawing, painting, etc are not going to be any less worth learning in the future due to AI.

Reply on news.ycombinator.com