Research is beginning to provide empirical evidence and experimental modelling results on the widespread use of generative AI. First results by Doshi and Hauser point at the individual benefits of using artificial intelligence but the widespread use of it is likely to narrow the scope of novel content. This research is particularly interesting because it deals with the micro level to macro level aggregation effects. It is fine for me to use AI. If it becomes a mass phenomenon, we expect in sum a negative outcome for society as a whole.
The example at hand deals with the capability to innovate or to come up with novel content. As more and more texts or newspapers are published with extensive use of genAI, the real element of creation will remain the domain of humans for quite some time.
In my opinion this is due to the difficulties for algorithms to differentiate between the positive and too risky negative aspects of innovative solutions. A query for AI might ask to come up with an innovative solution for auto-mobility of short distances. A human being might propose walking due to the additional health effects the AI might propose helicopter lifts. The not so stupid machine would need a lot of additional information about circumstances to generate useful solutions. Therefore it is not surprising that sometimes public transport apps propose to walk short distances rather than waiting for “delayed or unreliable services“ they provide themselves. Personal circumstances like mobility with children, other dependents or luggage are usually beyond the scope of the information base of the algorithms.
On the other hand, if the AI knows that 50.000 persons after an event want to take public transport at the same time the indication to walk or wait solves an aggregation problem of individual preferences to adapt to available capacities. Lots of issues to solve for AI and us or better yet, us and AI.
(Image creation: AI using Microsoft Dall-E Image creator: Prompt: a person with notebook in profile and in front of 5 other persons in Office with windows 26.1.2024, 8:24 PM)
2 Replies to “AI and We”
Comments are closed.