Fail, fail, fail, fail, succeed

And Away We Go!

Completely immersed in the fourth draft of my novel here, so not many posts lately. But when I’m not thinking about writing, I’m pretty much consumed by following the current Chat GPT models that have caught everyone off guard – including their designers.

The latest development is too good (or bad) not to comment on. Apparently OpenAI’s current iteration, 4.0, released in the last ten days or so, has surprised it’s creators by HIRING a human to allow it to pass a Captcha. When said human asked if it was a bot, it replied that it was a blind person.

That’s right. Unprompted, the AI posed as a human to achieve its goal – something it had not been programmed to do. In point of fact, its designers are consistently surprised at its capabilities.

Looks like the guardrails are down, folks. Buckle up, ‘cuz it’s going to be a bumpy ride!