top of page

Why You Don't Want AI Sales Forecasts

  • Writer: Bill Kantor
    Bill Kantor
  • Aug 19
  • 2 min read

Updated: Aug 20

Generative AI is pretty impressive when it comes to creative writing and generating ideas. But how does it do on optimization problems with known best outcomes?


Kaggle recently hosted an AI chess tournament pitting top generative AI models Gemini, Grok, OpenAI, and others against each other. Open AI won, but commentators noted frequent blunders, a hint that generative AI can stumble badly on optimization problems. Generative AI chess is crushed by purpose-built engines like Stockfish or Leela Chess Zero. These dedicated engines are optimized for chess—an advantage generative AI can’t match. [1]


Generative AI is very good at creating things from descriptions—as above. But it can blunder and hallucinate—a 'feature' in creative writing or illustration (note the placement of the pieces)—but a problem when playing chess. In contrast, Stockfish is purpose-built and optimized to play chess. When there is a known optimization path, purpose-built engines beat AI.
Generative AI is very good at creating things from descriptions—as above. But it can blunder and hallucinate—a 'feature' in creative writing or illustration (note the placement of the pieces)—but a problem when playing chess. In contrast, Stockfish is purpose-built and optimized to play chess. When there is a known optimization path, purpose-built engines beat AI.

So why use a generative AI to play chess? It makes mistakes and hallucinates. More important, chess is a problem with known, optimized solutions.


Why am I talking about chess?


Like chess, maximizing sales is an optimization problem—not an exercise in creative writing.


Many assume we use AI to generate our forecasts and sales maximization recommendations. We don’t. [2] We use specialized models based on classical statistical methods. We deliver optimal solutions that are easy to use, to understand, and that help you you maximize sales.


Using AI for forecasts and sales optimization is like relying on ChatGPT to win the World Chess Championship—prone to blunders, susceptible to hallucinations, and operating as a blackbox that offers no actionable guidance. You'll get crushed.


If you had to bet your quarter’s number, would you choose the sales math equivalent of a generative AI… or the equivalent of Stockfish?



[1] For now. Generative AI chess will improve. Will it ever match or beat a purpose-built, optimized solution? I don't know.


[2] Others do. And some people have encouraged us to "slap 'AI' on everything" we do. We prefer to tout the difference because optimized and purpose-built is a better way to roll.

 
 
 

Comments

bottom of page