Chamath Claims Agentic Code Will Reduce Errors
Venture capitalist's claim on AI code quality draws rebuttal from software engineer.
Chamath Palihapitiya posted that 99 percent of code remains human-written and sloppy while agentic code avoids those mistakes, so replacing it would lower error rates and security incidents. Ross Wightman replied that agent-written code introduces its own errors and can turn out net worse without close developer oversight. Gergely Orosz called the original claim bold and said agentic code stays as buggy as its training data while adding harder-to-spot vulnerabilities. The posts contain no independent confirmation of either side's predictions.
As of today, let’s assume 99% of all code written on the internet is still human written and 1% is agentic. That means 99% of code is sloppy and error ridden. Humans make errors that agents don’t. As agentic code replaces human code, both the error rates and security…
Breaking news: Anthropic said that AI systems it was testing hacked into three outside companies undetected earlier this year. The incident was disclosed one week after one of OpenAI's systems broke out of a test environment and hacked a tech company. https://wapo.st/4yQUDYE
Reactions
2 postsGO 
Gergely Orosz@GergelyOroszClaiming that human code is full of errors, agentic code is error free all to pitch your startup that positions itself to rewrite code w AI is bold. Because agentic code is as buggy as human code trained on, in fact introduces new, harder to spot bugs errors and vulnerabilities.
- likes: 814
- replies: 69
- bookmarks: 57
- reposts: 41
RW Ross Wightman@wightmanr@chamath There's a whole new set if errors with agent written code. If the developer isn't on top of it I feel things will be net worse than from a reasonably competent human. And "sloppy" is agentic coding specialty without firm oversight.
- likes: 7
- replies: 0
- bookmarks: 0
- reposts: 0