That video showed him saying that it’s good for autocomplete. But speaking from experience testing it on Rust, Python, JS, HTML and CSS, it performed the worst on Rust. It wrote tests well, but sucked at features or refactoring. Whether the problem is between the chair and the screen, I don’t know.
Whether AI will be able to write secure code, I dunno, I haven’t tried. It could be put into the rules to consider security and add tests relating to security or add an adversarial agent that tries to find flaws in the code which can be exploited. That could probably do more than a developer who has no time assigned to care about testing, much less security.
What it does to the IT sector in the long run - who knows…
Agreed. Things are moving so quickly, it’s impossible to predict. There are lots of people on LinkedIn screaming about obsoletion of humans or other bold claims, but to me they are like drunk fortune tellers: tell enough fortunes and one is bound to be right.
My naive hope is that local models or maybe workplace distributed clusters catch up and the cloud based bubble bursts. I am under the impression, that atm. a big difference as to whether a tool works well or not is more related to how well all the software around the actual llm is constructed. E.g. for discovery - being able to quickly ingest an internet url and access a web index is a big force of the cloud based providers atm. And for coding it has a lot to do with quickly searching and finding the relevant parts of the codebase and evaluate whether the llm has all the required information to correctly perform a task.
That video showed him saying that it’s good for autocomplete. But speaking from experience testing it on Rust, Python, JS, HTML and CSS, it performed the worst on Rust. It wrote tests well, but sucked at features or refactoring. Whether the problem is between the chair and the screen, I don’t know.
Whether AI will be able to write secure code, I dunno, I haven’t tried. It could be put into the rules to consider security and add tests relating to security or add an adversarial agent that tries to find flaws in the code which can be exploited. That could probably do more than a developer who has no time assigned to care about testing, much less security.
Agreed. Things are moving so quickly, it’s impossible to predict. There are lots of people on LinkedIn screaming about obsoletion of humans or other bold claims, but to me they are like drunk fortune tellers: tell enough fortunes and one is bound to be right.
My naive hope is that local models or maybe workplace distributed clusters catch up and the cloud based bubble bursts. I am under the impression, that atm. a big difference as to whether a tool works well or not is more related to how well all the software around the actual llm is constructed. E.g. for discovery - being able to quickly ingest an internet url and access a web index is a big force of the cloud based providers atm. And for coding it has a lot to do with quickly searching and finding the relevant parts of the codebase and evaluate whether the llm has all the required information to correctly perform a task.