Using AI demands expertise

Generative AI is NOT intelligent. It is a machine that uses fancy math to turn words into numbers and numbers into sentences. It appears “smart” in the same way a cloud can look like a human face. Moreover, GenAI does not “hallucinate.” Sometimes, the output of turning math into sentences does not conform to observed reality, but all the AI knows is that the numbers correlated.

That is not to defame AI, necessarily. In the long term, like many of the innovations that came before, it may grow up and demonstrate a societal value that lives up to or exceeds the initial hype.

But the history of capitalism is the history of financial and cultural manias. And the early founders in these morality plays are often greedy or corrupt, and their early investors are often both perpetrator and victim. Consider some earlier innovations-turned-scandals-or-misadventures that caused short-term economic harms but longer-term benefits:

• Tulips are an $8 billion global market.
• Railroad lines are still critical infrastructure.
• A global fiber optic backbone now connects the internet.
• Many dot-com startups crashed and burned so others could run.

Unfortunately, a possibly unique challenge with the AI hype cycle is the way it is such a direct-to-consumer product that also directly flatters the human brain’s inherent cognitive biases. From anthropomorphism to pareidolia (see ‘clouds’ in the first paragraph) and beyond, we are primed to accept output from a tool that "acts" human and lives in the same silicon box as Google and Wikipedia.

But GenAI is not Wikipedia (and Google itself is increasingly AI Slop). And to effectively use any GenAI tool at the moment, and maybe forever in any professional context, you need to be an expert in one or more of these categories:

1. An expert in the technology, knowing which platform is best for your specific needs, how to write the best prompt, and how to verify the quality of the output. OR...
2. An expert in the domain/topic and willing to evaluate the output and “negotiate” with the tool until you get or revise a suitable response. OR...
3. An expert in the suitability of the output for a specific purpose. If your application letter to Harvard works, or the requested meme is funny and shareable, or the vibe code works - that is good enough in some cases.

AI is just a tool. It can not replace "experts," though that is different than saying "experts" will not be displaced (see "capitalism" above). To use any tool you need to know what it is good for and not good for. And what the ethical/unethical applications of it are. So far, some ostensibly intelligent people are failing that test.