If you’re not aware of what it’s good at, given what very smart people are saying and doing with it, I think you’re either not paying attention or aren’t being intellectually honest with yourself
Instead of appealing to authority you could have given direct examples of how it's transformed your ways of working, that could've continued the conversation somewhere.
Or those people aren't actually very smart, or they're caught up in the hype, or since they are very smart they exist in a mode where their experience doesn't translate to normal, everyday situations.
It seems that AI coding tools are very sensitive to codebase structure. If you work on a monolith with relatively simple, straightforward structure this is the happy path. A bird's nest of microservices is not. If your team has taken the time and effort to structure the codebase in a way that's amenable to AI, and you invest in the tooling, and you keep up that effort over time, then AI does seem to work.. Not "10x productivity gain" as they try to sell it to us, but maybe >1.0x. It's not clear, though, that for the vast majority of developers AI provides any speedup whatsoever. That's the problem. If it only works for the top 5% or whatever, that addressable market is very, very small.
I've seen a lot of very rich people* say it's amazing, it's changing my life, it's going to change your lives (it's going to take away all your jobs so we don't have to pay you anymore), we're about to hit the singularity and start a new golden age with it.
I've seen some apparently-smart people say they're using it for all kinds of things and it's doing great for them.
I've seen roughly the same number of apparently-smart people say they've tried it, they've given it a really good shot, but it doesn't work well for them, and in fact, when they tried, it made them less productive.
When I've personally tried it (almost exclusively on local generation), I've found it entertaining, but not reliable enough to use for more than that. And I do not trust any of the hosted models not to take everything I feed them and monetize it, including by selling it to organizations like ICE which I find utterly reprehensible.
So while I'm not bigstrat2003, about me, at least, you're wrong: I am paying attention, and I'm being intellectually honest. I'm also evaluating it for more than just "does this make me more money in the short term?"
* Who just so happen to be heavily invested in AI companies...
Using nano banana does not require arcane prompt engineering.
People who have not learnt image prompt engineering probably didn't miss anything.
The irony of prompt engineering is that models are good at generating prompts.
Future tools will almost certainly simply “improve” you naive prompt before passing it to the model.
Claude already does this for code. Id be amazed if nano banana doesnt.
People who invested in learning prompt engineering probably picked up useful skills for building ai tools but not for using next gen ai tools other people make.
Its not wasted effort; its just increasingly irrelevant to people doing day-to-day BAU work.
If the api prevents you from passing a raw prompt to the model, prompt engineering at that level isnt just unnecessary; its irrelevant. Your prompt will be transformed into an unknown internal prompt before hitting the model.
> Claude already does this for code. Id be amazed if nano banana doesnt.
Nano Banana is actually a reasoning model so yeah it kinda does, but not in the way one might assume. If you use the api you can dump the text part and it's usually huge (and therefore expensive, which is one drawback of it. It can even have "imagery thinking" process...!)
Once you begin to see the “model” as only part of the stack, you begin to realize that you can draw the line of the system to include the user as well.
> the user inclusion part is real too. the best results i get aren't from fully autonomous agents, they're from tight human-in-the-loop cycles where i'm steering in real time. the model does the heavy lifting, i do the architectural decisions and error correction. feels more like pair programming than automation.
Precisely. This is why I use Zed and the Zed Agent. It's near-unparalleled for live, mind-meld pair programming with an agent, thanks to CRDTs, DeltaDB, etc. I can elaborate if anyone is interested.
The special (or at least new to me) things about Zed (when you use it with the built-in agent, instead of one of the ones available through ACP) basically boil down to the fact that it's a hyper advanced CRDT-based collaborative editor, that's meant for live pair programming in the same file, so it can just treat agents like another collaborator.
1. the diffs from the agent just show up in the regular file you were editing, you're not forced to use a special completion model, or view the changes in a special temporary staging mode or different window.
2. you can continue to edit the exact same source code without accepting or rejecting the changes, even in the same places, and nothing breaks — the diffs still look right, and doing an accept or reject Just Works afterwards.
3. you can accept or reject changes piecemeal, and the model doesn't get confused by this at all and have to go "oh wait, the file was/wasn't changed, let me re-read..." or whatever.
4. Even though you haven't accepted the changes, the model can continue to make new ones, since they're stored as branches in the CRDT, so you can have it iterate on its suggestions before you accept them, without forcing it to start completely over either (it sees the file as if its changes were accepted)
5. Moreover, the actual files on disk are in the state it suggests, meaning you can compile, fuzz, test, run, etc to see what it's proposed changes do before accepting them
6. you can click a follow button and see which files it has open, where it's looking in them, and watch as it edits the text, like you're following a dude in Dwarf Fortress. This means you can very quickly know what it's working on and when, correct it, or hop in to work on the same file it is.
7. It can actually go back and edit the same place multiple times as part of a thinking chain, or even as part of the same edit, which has some pretty cool implications for final code-quality, because of the fact that it can iterate on its suggestion before you accept it, as well as point (9) below
8. It streams its code diffs, instead of hanging and then producing them as a single gigantic tool call. Seeing it edit the text live, instead of having to wait for a final complete diff to come through that you either accept or reject, is a huge boon for iteration time compared to e.g. ClaudeCode, because you can stop and correct it mid way, and also read as it goes so you're more in lockstep with what's happening.
9. Crucially, because the text it's suggesting is actually in the buffer at all times, you can see LSP, tree-sitter, and linter feedback, all inline and live as it writes code; and as soon as it's done an edit, it can see those diagnostics too — so it can actually iterate on what it's doing with feedback before you accept anything, while it is in the process of doing a series of changes, instead of you having to accept the whole diff to see what the LSP says
Books by Peter Bevelin (From Darwin to Munger etc.) or Rolf Dobelli are decent compilations. But mental models are everywhere. Taleb's books have a bunch. But start with what you have in front of you: pick one and actually apply it programmatically, then add to your repertoire one at a time.
GenAI changes the dynamics of information systems so fundamentally that our entire notion of intellectual property is being upended.
Copyright was predicated on the notion that ideas and styles can not be protected, but that explicit expressive works can. For example, a recipe can't be protected, but the story you wrap around it that tells how your grandma used to make it would be.
LLMs are particularly challenging to wrangle with because they perform language alchemy. They can (and do) re-express the core ideas, styles, themes, etc. without violating copyright.
People deem this 'theft' and 'stealing' because they are trying to reconcile the myth of intellectual property with reality, and are also simultaneously sensing the economic ladder being pulled up by elites who are watching and gaming the geopolitical world disorder.
There will be a new system of value capture that content creators need to position for, which is to be seen as a more valuable source of high quality materials than an LLM, serving a specific market, and effectively acquiring attention to owned properties and products.
It will not be pay-per-crawl. Or pay-per-use. It will be an attention game, just like everything in the modern economy.
Attention is the only way you can monetize information.
No. The idea-expression dichotomy is a common myth about copyright law, right up there with "if I already own the physical cartridge, downloading this game ROM is OK".
The ONLY things that matter when determining whether copyright was infringed are "access" and "substantial similarity". The first refers to whether the alleged infringer did, or had a reasonable opportunity to, view the copyrighted work. The second is more vague and open-ended. But if these two, alone, can be established in court, then absent a fair use or other defense (for example, all of the ways in which your work is "substantially similar" to the infringed work are public domain), you are infringing. Period. End of story.
The Tetris Company, for example, owns the idea of falling-tetromino puzzle video games. If you develop and release such a game, they will sue you and they will win. They have won in the past and they can retain Boies-tier lawyers to litigate a small crater where you once stood if need be. In fact, the ruling in the Tetris vs. Xio case means that look-and-feel copyrights, thought dead after Apple v. Microsoft and Lotus v. Borland, are now back on the table.
It's not like this is even terribly new. Atari, license holders to Pac-Man on game consoles at the time, sued Philips over the release of K.C. Munchkin! on their rival console, the Magnavox Odyssey 2. Munchkin didn't look like Pac-Man. The monsters didn't look like the ghosts from Pac-Man. The mazes and some of the game mechanics were significantly different. Yet, the judge ruled that because it featured an "eater" who ate dots and avoided enemies in a maze, and sometimes had the opportunity to eat the enemies, K.C. Munchkin! infringed on the copyrights to Pac-Man. The ideas used in Pac-Man were novel enough to be eligible for copyright protection.
This is one of those fun "achsully" responses I get the privilege to refute.
It's a foundational principle of copyright law, codified in 17 U.S.C. § 102(b): "In no case does copyright protection for an original work of authorship extend to any idea, procedure, process, system, method of operation, concept, principle, or discovery"
Now, we can quibble over what qualifies there, but the dichotomy itself is pretty clear.
This goes back to Baker v. Selden (1879) and remains bedrock copyright doctrine.
The Tetris case is overstated. Tetris v. Xio did not establish that The Tetris Company "owns the idea of falling-tetromino puzzle video games." The court explicitly applied the idea-expression dichotomy and found Xio copied specific expressive choices (exact dimensions, specific visual style, particular piece colors). Many Tetris-like games exist legally, and it is the specific expressive elements that were considered in the Xio case.
K.C. Munchkin is old and criticized. That 1982 ruling predates major developments like Computer Associates v. Altai, which established more rigorous methods for filtering out unprotectable elements. The Munchkin decision continues to be debated.
"Substantial similarity" analysis itself incorporates idea-expression filtering. Courts use tests specifically designed to separate protectable expression from unprotectable ideas, especially when considering the four factors of fair use (when applied as a defense.)
I think what you'll find is that most aren't happy with the current copyright law anyway (I include myself in that group) or don't understand it or don't agree with it, and thus will just shrug.
For example, copyright duration is far longer than most people think (life of author plus seventy (or plus ninety-five years if corporation). Corporations treat copyright as a way to create moats for themselves and freeze competitors than as a creative endeavor. Most creative works earn little to nothing anyway, while a tiny minority generate the most revenue. And it's not easy to get a copyright or atleast percieved to be easy, so again it incentivises those that can afford lawyers to navigate the legal environment. Also, enforcement of copyright law requires surviellance and censorship.
Truthfully I think there will be a time when people will look at current copyright law the same way we now look at guilds in the middleages.
It's incoherent to be anti-copyright because it's used to freeze out competition by corporations and be pro-AI (which is exactly that, at vastly greater scale).
| Long-term value: We do not optimize for time spent in ChatGPT. We prioritize user trust and user experience over revenue.
The unspoken part -- This holds true so long as revenue is at least equal to costs, and speaks nothing about whether user trust and user experience is optimized over profit.
reply