I'll get scared when a chat bot can log into JIRA, take a ticket, read the requirements, say "Hmm, there's not enough details in here and a couple things seem ambiguous or contradictory..." Then it schedules a meeting with the biz dev and a QA resource and they all hash out and improve the ticket. Then it goes and implements the code and puts it out for review, addresses any comments, works with QA to clarify how to test it (maybe that's another chat bot) and deploys it.

Not to mention, until on-prem chatGPT is available, companies that care about source code secrecy aren’t going to expose their codebase to OpenAI.

We’ll get another slew of decades long court cases once one line of Oracle or Microsoft code proprietary code comes out of chatGPT.

We will be all dead and gone long before any of those points are a reality...