January 18, 2024

AGI or Stochastic Parrot? What To Expect From Generative AI in 2024

AGI or Stochastic Parrot? What To Expect From Generative AI in 2024

It's hard to argue that generative AI hasn't hit the mainstream now. ChatGPT and all the hype that came with it dominated the media and popular culture throughout 2023. But despite the buzz around gen AI, its mainstream adoption within organizations is still nascent.

The integration and real world use of generative AI in businesses' day to day operations has a long way to go. MIT Technology Review’s research reveals that as few as 9% of businesses have productionized any kind of generative AI implementation.

Perhaps more interestingly though, this Weka study reports that at least half of the businesses that did try rolling out gen AI failed. As obvious as the low hanging fruit looks for generative AI projects, the fact is that point solutions - papering over existing processes with a new system - rarely show real ROI and more often just bolster resistance to real change.

Sound familiar?

Implementing organizational change is hard. Always has been. And rolling out new technology without a wider strategy and a competent team usually comes with novel risk reward properties. Let's put it like that.

There are plenty of simple use cases for generative AI which are relatively low risk to implement right now. Document summarization apps, Q&A chatbots on your own data, and all kinds of apps that generate content and even running code within minutes. You want ideas? We've got lots of them.

However, we think that businesses eager to seize the massive first mover opportunities in AI will have to revert to more strategic and methodical change management practices in 2024. As larger companies drive demand more specialist models will appear as well as frameworks and best practices for deploying them. In the gen AI space this is probably an opportunity for anyone fine tuning domain specific large language models, or those building tools that help integrate AI at a deeper level in the business.

The real opportunities for AI lie in redesigning systems of interdependent decisions; not training a parrot to write your emails. The best companies will look for opportunities to grow their business while remaining focused on factors like security and operational risk.