In light of a number of reports about how AI use can make us “dumber,” including a recent study by MIT, I thought it might be useful to set out a few brief thoughts.
While I may have some ideas about the question of whether using AI may make us stupider, I would start with a somewhat different observation — using AI is itself stupid (and to be clear, I am using “AI” as shorthand for text/image producing generative AI models, not all forms of machine learning). It’s stupid because it valorizes the destination while eliding the journey. It’s stupid because current AI models are rooted in massive exploitation of our fellow human beings — from the cultural workers whose expression is misappropriated to train AI models, to all of us whose personal information is repurposed to create bespoke/unshared realities and to sell us goods and services, to the underpaid “moderation workers” who toil in intolerable conditions, forced to spend their days sifting through the worst detritus of human expression in a vain attempt to sanitize a commercial product.
And that’s before we even get to the issues of environmental destruction through huge consumption of energy and water involved in creating “efficiencies” of output that manage to drown truth, beauty and poetry in irrelevance, both by design and inadvertently. In short — AI, at least in its present incarnation, is a deeply toxic tree, the fruit of which is poison to our minds and dulls our senses, replacing originality with derivative output based on predictability. It is, almost by definition, a stupidity-producing machine.
As Dr. Heidy Khlaaf tweeted earlier today: “We live in a time of mass hysteria but it’s not controversial to prioritise human lives over training AI to carry out basic tasks humans are already capable of doing.” So maybe we should heed her counsel? Let that be our guide.
