Gary N. Smith, who writes frequently for Mind Matters News, was interviewed for a recent article from The Daily Beast on the developing AI “arms race” among tech giants like Microsoft and Google. Because of the success of OpenAI’s ChatGPT, tech companies are scrambling to adjust and compete.

Microsoft announced that it is now incorporating ChatGPT into its search engine, and Google advertised its AI alternative “Bard,” which they also plan to integrate with their browser. This comes just a couple of months after the revelation of ChatGPT-3, which saw several million users within the first week of its existence. Smith commented,

“Big companies don’t want to miss the next big thing, and startups want to cash in on unrestrained hopes and dreams, the way so many startups cashed in on dot coms, cryptocurrencies, and generic AI.”

Google and Microsoft’s Chatbot War Might Change the Internet Forever (thedailybeast.com)

Smith thinks that the confidence these companies are demonstrating toward Large Language Models (LLMs) is naïve, and there could be consequences. He continues in the interview,

“LLMs should only be used in situations where the costs of mistakes are small—like recommending movies—but their magical powers will surely persuade many people that LLMs can be used in situations where the costs of mistakes are large, like hiring decisions, loan approvals, prison sentences, medical diagnoses, and military strategy.”

Because LLMs can’t understand the meaning of the words they generate, failing to separate fact from fiction, they shouldn’t be trusted with big roles and choices.

In addition, AI, because it’s based on biased data sets, can only generate biased responses. As Richard W. Stevens notes here, its biases seemed to be radically skewed politically.

Smith believes AI is limited ultimately because it “writes before it thinks.” Because it lacks an understanding of the meaning and implications of the words it generates, it should be used more cautiously. But, big companies like Microsoft and Google are jumping the gun and going with the next big thing, which is hardly surprising.