AI and complements
In a “weak links” model of production, where tasks are complements, the gains from automation tend to arrive slowly because output is constrained by the least productive tasks. But the same complementarity makes attacks more harmful and easier to execute. For example, with an elasticity of substitution across tasks of 0.2, automating tasks worth 1 […] The post AI and complements appeared first on…
In a "weak links" model of production, tasks that are complements tend to yield gradual gains from automation, as output is limited by the least productive tasks. However, this same complementarity makes attacks more damaging and easier to carry out. For instance, if the elasticity of substitution across tasks is 0.2, automating tasks that account for 1 percent of GDP and increasing their productivity tenfold only raises output by less than one percent, whereas reducing their productivity tenfold results in a 68 percent decrease in output.
When an expanding AI sector simultaneously enhances AI models' capabilities and increases the likelihood of attacks, the beneficial productivity spillover is minimal compared to the potential harm, leading markets to overinvest in AI. In a weak-link economy, the costs and benefits of AI can be extremely asymmetrical, as noted in a recent "in progress" manuscript by Charles I. Jones and Christopher Tonetti, reported by Lukas Freund.
Written by urgent.news from Marginal Revolution's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
