When AI rights become a big issue, AI might be automating lots of jobs. Does that matter?

If society is debating granting moral status to AIs, the same AIs might be automating a lot of jobs. There are several ways this could influence attitudes on how to treat AI. If people feel economically threatened by AI, they could be particularly hostile towards anything that looks like giving AI more power. 

One indicator could be the effects of past automation on attitudes towards outgroups. There is a literature (not unanimously) claiming that losers of automation turn right-wing and anti-immigrant. In my  MSc Dissertation, I asked whether automation might affect moral universalism in general: If robots are taking jobs in your industry, do you generally respect outgroups less? 

The answer appears to be no, at least in my data. In Germany between 2002 and 2018, automation-exposed workers didn’t turn relatively more or less morally universalist in the European Social Survey Data. Of course, this won’t straightforwardly translate to future AI automation. In particular, German automation was benign and created many new jobs. But AI automation might assume a similar shape - at least, current labour market effects are modest. 

Separately, the issue of economic shocks and outgroup treatments holds another lesson for digital minds advocates. One theory explaining hostility to immigrants assumes that as salient outgroups, they are simply convenient scapegoats and outlets of frustration, even if that frustration might have other economic sources. In the future, AIs themselves might be a salient outgroup whose proper place is fiercely debated. As such, they might become an outlet for frustration, regardless of their actual economic impact. For reference, Americans currently lean negative in their views on AI’s economic impact, even as the AI buildout significantly contributes to US growth. 

In sum, AI’s economic impact probably plays a minor role for AI welfare where AI’s effects are modest or ambiguous, because automation doesn’t appear to influence moral universalism at these margins and AI’s impact is not correctly attributed anyway. But this might change for drastic, obvious impacts on the economy and society.