ajjenkins

It looks like they had a hard time with the magic wand, peacock-to-cat example. Despite using a very long prompt that repeatedly stresses that the video should look photorealistic and NOT like a video game, the output video looks like a video game. The model is clearly capable of generating photorealistic humans and animals, so I wonder why it struggled to do so with that particular prompt.

Interesting. It actually moved the ball next to the elephant, but I guess that accomplishes the goal of the elephant and ball being next to each other. But probably not how must humans would interpret that prompt. And it continued to move both the elephant and ball after that for some reason.

Hopefully this is like the first GPT models and it gets better over time.

This is fun, and a good demonstration of Jev’s capabilities.

It’s obviously much faster than a human, but it’s interesting that it only got 7/10 right on both rounds (same as me). I’m wondering what Jev could be trusted to do reliably given these results.

I guess it wasn’t optimized for these types of trivia questions. Or maybe the image capabilities are too new and need refinement.

In my opinion the the point of the site is that Jev has knowledge on all topics that makes it hard to beat. It might not be as smart as you, but it probably has a greater breadth of knowledge.

I was confused by the chess board question too. But I do think the question is worded correctly—it’s just a little unintuitive.

Much better now, thanks!

I would describe the website as “twitter for developers and uses your GitHub account for auth”. It’s a simple concept, but I think using your GitHub account to sign in is a great idea and a meaningful differentiator. Good luck!

It would be nice if the website had more information about the platform before you create an account.

They also found stone anchors. So you could say it’s a shipwreck where nothing is left but the anchors.

If the anchors were gone too, I would agree with you—some piece of the original ship (not just the cargo) must remain to consider it a shipwreck.

A lot of people in the comments say that the supposedly better examples in the article are still “horrible” because they can tell they’re AI and point to things that the AI did “wrong” and a human artist would do better.

That may be true if you’re working with a top of the line artist, but in my experience the average freelance designer online is significantly worse than AI. I’ve worked with dozens of budget friendly designers on Fiverr and consistently get way worse results than I’ve gotten from AI models. Will a great human designer produce better output than an AI? Maybe, but a small local event or business probably won’t have the budget to hire them or know where to find them. It’s cheaper, faster and easier to use AI to generate something that’s good enough for their purposes.

People also point out all the little flaws in the “good” examples and cite them as obvious tells that they were AI generated. I think it would be interesting to see if people could reliably distinguish between AI generated flyers and human made flyers. Sure, sometimes it’s obvious like when the AI adds those weird, cartoony people. But if the model is told not to create designs like that and stick to the styles used in the second half of the article, I expect people would have a much harder time distinguishing the human designs from the AI designs.

This assumes a future where everyone is an entrepreneur, self-starter, creator or innovator. But people with the drive to create their own ideas and value account for a very small percentage of the population. Most people just want a steady paying job that they can perform well (or not work at all). I’m not sure how “putting AI in the hands of everyone” will help this part of the population. I feel like this is a really tone deaf take from Meta that doesn’t address people’s concerns around AI.

In deductive puzzle games like this, it’s nice when the hints teach you how to play the game instead of just solving the puzzle for you.

I got stuck and couldn’t figure out what to do next. Maybe as a first clue, you can highlight the next cell they should be able to figure something out about (either clear it or flag it). If they’re still stuck, then you can clear it or flag it.

Even better would be to explain the logic for why it’s safe to clear that cell or why it should be flagged, but I know that’s harder to implement.

The game is well made and has a lot of potential. Great work!

Another good feature in ColorSweeper that would be nice in your game is a “sketch” view that lets you place potential mines/fangs on the board so you can see how things would play out. Like if you put a 1-fang here and a 2-fang there, does that work with all the numbers shown.

This is fun. But I think you need a way to specify in the flag how many fangs are there. A similar modern minesweeper game is ColorSweeper.

A feature I like in that game is it will grey out numbers once you’ve sufficiently annotated the adjacent spaces.

It’s a really good game, but you need to pay $5 to get rid of ads.