AI’s silly season

We are in the silly season as people try to define AI’s impact without having much of a clue. There are the crypto people who have seamlessly become AI people, much like how the amateur Twitter epidemiologists suddenly morphed into military experts. I just want AI to help me navigate Excel before looking to construct an AI god to worship, but we all have different priorities. There’s a burgeoning market to playing the role of Toto, yanking back the curtain to reveal that the “Wizard of Oz” wasn’t all he was cracked up to be.

There’s money to be made by slapping the term “AI” on everything. This is inevitable. It’s the American way. I remember when “big data” was the moniker used by every assorted ad platform. After BuzzFeed saw its stock jump based on a passing reference to using AI to gin up lightweight quiz content, its lead has been followed by other publishers, who murmur about “looking at” using the tools or dabbling in using them.

This is a strange period for these tools. Like any new tech development, they are likely both overrated and underrated. The focus on hallucinations and misstatements obscures that these are research or beta versions. I’ve spoken to people who have seen demos of GPT-4, and the advances in this area appear to be coming quite fast. Who would expect the experience of ChatGPT now will be anything like using an AI chatbot in three years? I wouldn’t take that bet. And the shift in expectations for conversational interfaces will have lasting effects.

AI tools point the way to leveling the playing field between big publishers and small. Consider Semafor’s Gina Chua’s tinkering to use AI for copy editing and news editing. Are these tools there now to replace humans? No, but it’s obvious AI will make it possible for smaller publishers to more effectively compete with big publishers. Efficiency without scale. Of course, at big publishers, old habits die hard, so some executives see an opportunity to shake down AI for payments because publisher news content is being used to “train” these large language models. Get ready to litigate the scraping wars. Publishers will end up facing the difficult choice of whether to close off their content from these models, at the risk of disappearing from search engines.

The good news is humans continue to prove themselves adaptable. We can even sometimes beat the machines at their own games. The early hiccups will likely be forgotten. Let’s face it, the internet was dull for a long time, and a little chaos could reset the board and make it more interesting. Too much of the doomering willfully gives the impression these glorified auto-complete tools are sentient. They’re not. Instead, their uncanny ability to simulate human conversation will serve as a reminder of how important the real thing is, as Very Fine Day’s Brad Esposito put it: “The evolution of artificial intelligence is not a reason to be sad or angry or anxious — at least not for very long — because it is through this evolution that we will allow ourselves to truly focus on what matters.”

Συνέχεια εδώ