It’s intelligent! It’s revolutionary! It’s paradigm-shifting! It’s… marketing?

How much of the glitzy AI fanfare is justified and how much is just damn good PR?

As tools like ChatGPT dominate the public imagination, it’s easy to forget that they’re just really, really sophisticated auto-complete that learns from whatever’s available on the internet. And that material isn’t always great. 

Scientists predict that by 2026, AI chatbots will plateau because they’ll run out of quality, human-created training data—there are only so many Moby Dicks out there to lift from. When language models have nothing left but Tweets and 4chan comments to train on, their dazzling sonnet-writing abilities will start to dwindle—unless they re-read Melville. 

The much-hyped neural networks that form the basis of today’s AI tools might be artificial, but they’re not intelligent as we understand that term, bearing little resemblance to human brains. Yes, they can recognize patterns and can generalize, but they’re not capable of conceptual knowledge or deliberate reasoning. 

For now, when it comes to delivering measurable value, it’s less about the miracle of generative AI and more about the usefulness of ML (machine learning). Specifically, predictive analytics that can inform operational decisions. For example: if you can accurately predict when most customers will sign up for your service, you can get in front of the right audience that much faster. Is that modest and mundane? Yeah. Sorry.