As I was scrolling through the Financial Times (as one does), one article caught my eye: "Google and peers weigh an AI Prisoner's Dilemma". At this point in the semester, seeing the words "Prisoner" and "Dilemma" in any form together warrants my attention, and this article did not disappoint. The author argued that companies like Meta and Google are experiencing a prisoner's dilemma, because although the companies would be better off working together to develop AI, they still choose to compete separately. With trillions of dollars in potential value to be gained from AI, these companies want it all to themselves and thus compete and research AI separately.
While there certainly is a prisoner's dilemma regarding AI development, I think the article is missing another prisoner's dilemma that may occur in a few years. Although there certainly is real value to be had if firms work together to develop a strong functioning AI model, I believe another prisoner's dilemma may occur with pricing. If firms can collaborate on prices, they can charge much higher than with competition. With trillions predicted in a competitive market, the amount of money firms can make together would be astronomical. It remains to be seen if firms will go down this route, but if AI companies can band together, their profits will go through the roof.
1 comment:
Extending on prices, under a competitive market, firms would have to restrict the amount of AI products (quantity) they sell to raise prices. But Olson says the individual firm would only consider doing this if E_d (elasticity of demand) <= F_i (share of industry output). The benefit (higher price) has to be weighed against the cost (decreased output). It would not make sense for any firm to do this here because the AI market is highly competitive and growing.
Moreover, AI products are discrete (consumers either buy or don't) and there are network externalities: more users, more data, and within companies, increased productivity. But this means AI companies need an initial consumer base (accomplish by providing the good at zero to low cost), then price if their product becomes dominant. One concern with competition is partnerships between AI and large tech companies; if they control access to data and other resources then that will limit entry into the market.
Post a Comment