[dead]

Current models are amazing but production code incorporates unknowable information through decades of diagnosis and bug fixing (and sometimes the opposite: working around bugs in other software that it talks to). Put even the most amazing coder that was ever born onto improving a complex old codebase, and you can guarantee they are going to break stuff. AI is better but not infinitely better.

These are good questions that we won’t get answers to anytime soon.

I’d love to know what internal services and tools that are included in this, as well as any deliverables that shipped as software products to customers.

If it’s legacy code for IOS firmware, however, I’d be very worried.

I have zero doubt it could do it, if the prompter could competently explain the requirements. I have no reason to believe that this single engineer has any idea what problems most of the existing legacy code needs to solve or why it solves them that way...

It can create a UI that looks like Photoshop where half the features are broken and the other half are literally not even connected to anything, just copying menu options with no implementation to fool gullible people like you*

It can clone photoshop.. Have you considered why?

The ability to clean room Adobe products came at about the same time Claude's RE abilities suddenly improved.. Hmm lets think about this a little more.

Anthropic sucked the bones dry of the public domain, literature and all of opensource. It seems likely that Anthropic RL'd Claude into being a great reverse engineer because it needed more training data.

I believe that Anthropic turned Claude into a digital software shredder and started decomping proprietary code, like adobes.

So the reason Claude can suddenly clean room Photoshop isnt because it became a more capable programmer, but rather, the Adobe code is literally in its training data now.

Ciscos internal legacy code is not going to be in Claude's data (yet).

The decision to increase Claudes ability to RE is brilliant, because now everyone is doing the source code decomp for them, and sending them fresh data from previously closed source repositories of code.

lol