I was recently at a networking event when I said that I believed the AI bubble was about to burst. That statement needs unpacking – let me explain what I actually mean.
AI has actually been around for over 70 years – it didn’t arrive with ChatGPT but what did arrive in 2022 was the first version that felt accessible to everyone.
I’ll skip the debate about whether ChatGPT counts as ‘true’ AI – that’s a rabbit hole for another day.
Going back to the bubble bursting, we are already seeing signs of challenges in the AI space;
- US Government restricting access to the latest Claude AI model – story
- Ford rehiring workers that were replaced by AI – story
- Gizmodo are reporting that companies are hiring coders to fix AI screwups
- Companies are discovering that AI is more expensive than the employees it replaced – story
Four data points don’t make a trend but they do make a pattern worth paying attention to.
Now some of this can be put down to poor implementation of AI. However, with Governments getting involved with AI and potentially restricting access to the models, then it becomes less viable outside of the country of origin.
My belief is that the mass market take-up of AI will start to fall back with small to medium size businesses starting to come across challenges in implementing and using AI.
This will leave the big corporations the only ones with the skillset and money to implement AI successfully.
For small and medium businesses, the practical takeaway is this: be selective. AI tools that solve a specific, well-defined problem – writing first drafts, summarising documents, answering customer FAQs – will continue to be worth using. But wholesale replacing people or processes with AI, particularly where the output needs to be trusted, is looking increasingly risky. The companies that got burned weren’t using AI badly. They were using it everywhere.