Not to say there's no value to be had. However, it's also been oversold and its error-proneness is a real problem.

But, it seems there's a steady stream of commenters and creators who are quick to remind us that it's never a problem with the tech. It's always a user problem.

Meanwhile, the errors are renamed "hallucinations", but let's not focus on those. In fact, if you just look right over here at this .md file...

In any case, the point is that low quality code shouldn't be promoted as the acceptable goal for AI.

OTOH, if someone insists on writing bad code—even after being advised otherwise—then they should expect to be out of a job.

Doesn't seem like "terminated junior developer" is the level of quality we should be accepting or promoting for AI.

How are people getting this "waste of time not to use AI" and "generates 99% of my code" level of code quality?

I know people's response will likely be something, something, "add instructions to memory to not use n+1", etc. But if the Ai wants to generate code this bad, then it represents an overall quality risk that would require an infinite memory file.

So, none of this is noteworthy, but we're...noting it?

Not quibbling with you, personally. Quibbling more with the state of the hype.

>But September 2028? It might just be weird to run companies without AI management.

Coding is pattern matching and still requires a human in the loop to manage.

But, any human in the loop managing the "Ai management" is the manager, by definition.

Only way true Ai management is viable is via non LLM Ai, so completely depends on advancements there.

Serious question: why? Is it not just pointing it at its massive training corpus for a list of unsolved problems, and possibly even by degree of perceived difficulty? I'm trying to understand why finding the problem isn't a simple "search engine" style challenge, at which LLMs excel?

If so, I would think that would result in worsened performance, not quality (unless you are also suggesting they may be quantizing).

>Opus 4.8 and Opus 5 seems worse models than Opus 4.6

After all we've heard about benchmark cheating, I'm earnestly not sure which or whether benchmarks are reliable anymore. But, beyond the models, I wonder if changes to their harnesses and/or instructions dumb them down. I have noticed models change their behavior, even when using the same version/effort. Sometimes for better. Sometimes for worse.

And, I have noticed a model go from really good to struggling. On 4.8 things were going well for a good stretch, so I did not switch to 5 when it came out. Even after hearing complaints about 5, 4.8 was still going well. Then, suddenly over the last few days, 4.8 seems to have nosedived. It feels similar now to the complaints I hear about 5.

In my case, it suddenly started ignoring my design guide, and introducing new fonts etc. It would even use several different fonts and sizes, as well as different margins for similar elements within the same page. It abandoned classes and started inlining styles. It started feeling random and, even after it realized it needed to go back to the design guide, it just continued with more of the same.

There seems to be something that happens after new model releases in both quality and behavior of previous models. It may not be immediately, but eventually there is frequently some regression.

But, is it true?

>It means nothing...shallow...

Ironically, the entirety of your comment just repeats that the GP comment means nothing. There is no further explanation or "depth".

As it is, no one was discussing Europe, let alone holding it up as a model society.

What-aboutism is not a very compelling argument.

It's as if they are looking for an exacting duplicate of previous regimes before they become concerned—the agency name must be "Stasi" and uniforms must be of the same color.

-Mark Twain

I attempted to clarify that the impositions themselves are not deterministic, by indicating that the entire process is still probabilistic.

Maybe Anthropic's explanation is simple enough [0]:

>When watermarking is used, choices are still made at random, but the source of the randomness is different. Instead of using an arbitrary random number generator to pick the next word, watermarking uses the key and a few words that come before to settle what word the model should pick. That is, the words that Claude picks are still random, but now, one can check the sequence of words and see if it’s consistent with the choices Claude would make if it was using the key. If it is, one can assign a probability that the text was generated by Claude.

[0] https://www.anthropic.com/news/claude-text-watermark

>use it to store arbitrary information

No additional data is embedded. The range of available data is constrained by the text being generated (i.e. the sets of "next words" per text).

CC:

"The problem is that I overreached..."

[Wall of words here]

"Two things: window surface is limited. Extract template. Buffer result and add to surface. Then, follow-up with new model..."

Me:

What do you mean by "window surface" and what result are you referencing? Also, why do we need a new model?

CC:

"Ah, you're correct to point out that no new model is needed. The problem is elsewhere and once we address that, the existing model should work fine. Now, as to your question about..."

[Wall of words here]