Sector rotation, rate-related valuation compression and de-risking aside, US Tech stocks have become more volatile in the last 6 weeks for fundamental reasons as well. Whether it is Apple’s decision to not provide iPhone unit sales or Facebook’s data management challenges, the question of scalability is now part of the discussion around this group.
Fair enough, we say – outsized market caps are always a function of addressable markets and visible pathways to profit from them. Shrink either variable, and asset prices will decline. Make investors question both, and valuations can plummet.
The Pew Research Center was out recently with a survey about exactly this topic; the report describing the results is titled ”Public Attitudes Toward Computer Algorithms”. In our view, no issue better addresses how much runway Tech companies have left to leverage their data sets and the infrastructures that surround them. The entire investment thesis behind Google, Facebook, Amazon, Netflix, and many other Tech names centers on algorithmically scaling their businesses to address business/consumer needs and wants.
The results of the Pew Research survey of +4,500 respondents show the limits of what Americans think is acceptable in terms of algorithmic decision-making. Some examples:
- 58% thought “computer programs will always reflect some level of human bias”.
- 68% thought an algorithm that used non-financial metrics to determine access to credit was “unacceptable”. Worries over privacy were the most commonly cited concern.
- 67% thought it was “unacceptable” to use automated video analysis of job interviews and cited “removing the human element” as a chief concern.
- 57% didn’t like the thought of automated resume screening, both for the lack of human judgment and over concerns of fairness.
- 56% thought it was “unacceptable” to use algorithmic analysis to determine parole eligibility.
The funny thing about these responses: companies and governments already do all of them to some degree. Automated resume review is a widespread practice in HR. Pew mentions that “numerous firms” already offer non-traditional credit scoring. They also report that several multinational companies review interviews with AI-powered video analysis algorithms and numerous states across the US have rules based parole protocols.
Our take: Pew’s survey questions may be a bit specific, likely to elicit an emotional response, but they get to the nub of Tech’s problem. We already know Google and Facebook can sell advertising, Netflix can find you something to watch, and Amazon can help you buy the best vacuum cleaner. Companies are happy for the data behind the first two, and consumers see real utility in the latter duo.
But where do these companies go from here, and how fast will Americans let them go on that journey? The Pew survey does point out that younger respondents are more comfortable with algorithmic decision-making. But that means we’re talking about generational (i.e. 20 year) cycles. Not next year. Or the next decade. Until then, consumers will worry about fairness, objectivity, and other important issues.