I use Sonnet (4.5) every day and itâ(TM)s improving much faster than I expected. But I still correct it every 5 minutes. My first attempt to use Opus 4.6 as an agent failed immediately. Somehow Sonnet 4.5 did the same task better than Opus 4.6. But people are claiming fantastic results, so Iâ(TM)ll test it some more this week. Maybe Opus has a different âoepersonalityâ and requires a different style of prompt.
Last week I had to argue with my companyâ(TM)s customer support because our support AI hallucinated a feature that âoewasnâ(TM)t workingâ.
Most of the layoffs are strategic repositioning and not âoeAI automated their jobsâ scenarios. The CEOs are using AI as an excuse in some cases, but details of their layoff and hiring strategy tells a different story. There are some well researched reports about this on the internet if you care to look.
So a lot of this news frenzy about AI is hype. The AI companies are probably running a bunch of agents filling comment sections with fake testimonials to draw customers to their platforms. And the news media is trusting these CEOs and their PR departments.
AI is coming for our jobs soon. But itâ(TM)s really hard to tell when that will happen.