mpalmer

Well, I could argue that by inviting a comparison between trusting a single fact from Wikipedia and trusting arbitrary generated output, you are practicing less critical thinking than you otherwise might.

The example's not relevant; the two things are not analogous. Moreover your point only works if one accepts your conclusion that:

    Independent verification is the main way they measured critical thinking

And this isn't especially true. All this information was self-reported to begin with. Maybe what you're describing is a pattern in the responses where participants associated verification activities with critical thinking.

But they also reported assessing things about the content, like style/personality, logical soundness, relevance to a topic, or suitability for a specific purpose.

Studies don't need to justify themselves with concrete conclusions; gathering the data for future use is a fine goal.

I'm going to assume you're being serious.

Critical thinking is actually something that people exercise (or don't) all the time. It is a skill and practice that is inarguably valuable in all situations. This paper is not about the effects of critical thinking on "AI results", but the exact inverse of that.

Do you value the capacity for critical thinking in yourself? In others? If so, the effect reported by this paper is probably of interest to you.

Native already had very specific meaning to technologists. Seems to me the industry is trading on the existing credibility of "things that are native"

Downvote me if you want, I'm flaunting the guidelines in service of the discourse.

If I may posthumously discipline the horse - when did the suffixes "-native" and "-first" merge in unholy matrimony?

Was it when B2B marketing realized that engineers equated "native" with "best/fastest/simplest solution available"? Or is it even more stupid than that?

    You can't say this. We have no idea. There is nothing about (s/the law of physics/US history) that pushes (s/cancer cure/the right to privacy) a long time away.

Don't mind me, just repurposing one of your old comments to save time.

Me personally? I'd feel shame for not spending the half hour it would have taken to put even the flimsiest human stamp on the thing I asked an LLM to make for me.

The latest version of jj supports colocated git worktrees as part of its workspaces feature.

I don't think you can say that making wildly successful games one after the other is "feasible" for anyone who's interested in making games.

Challenge is never gone.

Here's a challenge: see yourself from the outside.

Realize that however effective you are as a programmer now, you were once a hopeless beginner.

Then you must acknowledge that you are still a hopeless beginner, relative to the you that doesn't exist yet.

In the spirit of fellowship, I tell you:

    git gud

---

And as you continue to learn, you'll see this for yourself, but I guarantee to you now that AI opened paths to writing novel software which no one has noticed yet. Software that was possible before but perhaps annoying to build, etc. Software which perhaps will only ever be imagined by someone who learned exactly the things you have, in exactly the order you did.

Never make the mistake that AI knows how to do everything just because it knows how to do everything we've done so far.

And don't think of your work as "mining" engineering problems! Forget engineering problems. Find a human problem and think about it for an entire month.

He had previously built and sold two games, each of which made him millions, funding the next effort. Not infeasible, but building on repeat success.

    My observation is that right now AI coding agents lean towards re-implementing everything from the ground up. This of course will not scale. It will scale beyond what humans writing code would scale to, but we will ask the AI for more, so it will even things out.

Those who cannot do, preach.

There's a funny trick where if you are an experienced software developer, and take time to think about your design and describe what you want in clear terms, it will actually do things your way. Weirdly, this still saves massive amounts of time.

People who don't understand how to build good software LOVE to expound on the things we'll have to design in order to make AI tools "lean toward" building better software.

It seems that these folks are not willing or able to conceive of, you know, actually getting good at this stuff.

For this reason, it never occurs to them that human beings are STILL the source of the best ideas and designs.

The evergreen observation; The most self-promote-y among us are mostly using AI to reinvent solutions and disguise them as new designs. These address solved problems, which are themselves disguised as new agentic problems.

Celebrating and promoting this stuff does not help us get better at discovering and building actually new things; change my mind.

I'm confused. Is this a paper or a blog post? It's basically a play by play of "something we asked an LLM to do". Is this what constitutes research at DeepMind these days?

You should play around more, it's flawed but more than any other tech it rewards exploration.

    But asking for something more nuanced, in the range of regular speech, requires the correct lingo

I like asking models to write things using only the N most common words in a language. It's probably never perfect, but it understands the assignment well enough. from there I notice things I prefer about that constraint, and things I don't.