What is happening to the AI Bubble
- Kevin van Heerden
- Sep 1
- 2 min read
By Steven Buckle - Conjurian
According to the Wall Street Journal, last week Meta’s Mark Zuckerberg put a freeze on AI hiring after a spending spree poaching top talent from OpenAI because the spiralling costs have drawn investor scrutiny.
Meanwhile, in Fortune Magazine an article by Jeremy Kahn reveals that a Massachusetts Institute of Technology (MIT) report into companies using AI shows that 95% of AI pilots fail and that the reason for the failures should make C-suite executives nervous.
Sam Altman CEO of OpenAI said “Are we in a phase where investors as a whole are overexcited about AI? In my opinion - Yes.” The MIT study analysed 300 public deployments, interviewed 150 leaders and surveyed 350 employees connected to recent AI integrations. Almost 40 billion in enterprise investment into generative AI and 95% of them had little or no impact on the bottom line despite wanting rapid revenue growth. They all tried to build their own AI tooling rather than using existing hardened ones. Companies that paid a third party were better off.
Five percent did not fail such as that by Eric Vaughn, CEO of Enterprise software IgniteTech, who is now experiencing a 75% profit margin increase since 2023. In Futurism’s article by Joe Wilkins, it was said about Eric that “CEO Boasts that he laid off 80 percent of his staff because they didn’t love AI enough” and that he threatens to do it again. Eric is quoted as saying “It was extremely difficult.”
The conclusion of the MIT study is that the AI models are smart enough but that the humans are not using them well enough. It is a skill issue, with brittle processes, lack of context, and misalignment with day-to-day operations contributing. Software developers will still have a job for the foreseeable future especially if their companies make sure that they are trained and listened to.
Key Points
All the failures were in Software Development projects
All failed to get some good external expertise
All tried to build their own tooling
All did not train their people thoroughly
Key Takeaways
Use Agentic AI for the other areas of your business
Get quality external expertise to help you
Use existing tooling
Pay real attention to your humans, your training, and your interconnected processes
To find out more, click the link below to book a meeting with us to see how we can help you.
Comments