Skip to main content

Overthrowing the CEO mystique: the robot boss

I’m a strong believer that the CEO space – that expensive, padded space that costs fortune 500 companies hundreds of millions per year – could be radically altered and made much less expensive by replacing CEOs with expert systems.
However, my faith in this program isn’t just based on the fact that generally, CEOs don’t provide much of an advantage to the firm – research consistently shows that CEOs who outperform do so in ways that undermine long term performance, and that the company often experiences crises and shock in the wake of CEO hotdogging, as the president of the company leaps to another post in another company. My faith is based in the improvement of expert systems.

A good study of the history of expert systems in law was published last year by Phillip Leith, who in the 90s was a strong critic of basing legal expert systems on Logical programming, under the ideological influence of Hart’s notion that the law can be reduced to rule-based behavior. It is a fascinating read. (The Rise and Fall of the Legal Expert System, in International Review of Law, Computers & Technology, 2016 Vol. 30, No. 3 – for those who want to look it up), and not just because Hart’s theories were put to an unexpected empirical test – not something that often happens in philosophy. It is also because the problem in setting up an expert system in law – how to represent “contextual” knowledge – is also at stake in building a management expert system.

Leith has a nice ability to compress an argument down to its essentials. His summary of what was happening in the 80s and 90s in AI is very deft:

“A relatively simple idea underpins the notion of a legal expert system: that one can take rules of law, mould them into a computer-based formal system, and advice will come out the other end. It was not uncommon to hear funders of research projects in the 1980s assert that to build a legal expert system, one had two basic and essentially simple options:
translate legislation (‘the law’) into some formalism and add a software interpreting mechanism as a front end for the user;
take a group of experts off for a few days and get them to lay out the relevant rules of law which can then be moulded into a formalism by a non-expert and, once again, add the interpreting user interface.
It is as if Occam’s Razor has been applied to the whole confusing business of ‘what is law’ and we are left with an elegant core notion which can be implemented by technicians. The model is thus of a core of rules, and a logical interpreter which parallels legal advice giving. This, I argue, was partly hubristic but is also a relatively accurate description of the non-critical perspectives around law schools during that decade. In fact, such a perspective still demonstrates its attraction to the technician and research funder (The European JURIX community has continued to publish in this research spirit). The promise being made in the 1980s was that cheap, good quality advice would allow us to discard the need for expensive experts or leverage their productivity further than could the traditional ‘fee earner’ basis.”

Leith’s story is, in part, the story of the hyped futurism of the 90s. However, artificial intelligence and expert systems have certainly moved on, tackling just the procedural and representational problems he is talking about. No rule based computer system will take over the upper management position. The recent speech by the head of Alibabi in China, Jack Ma, who predicted that robots would take over from CEOs because they have no emotions, is precisely wrong. Jack Ma’s speech is, in fact, a back to the future creed that must have made AI folks groan.

In fact, unemotional robots would make suck CEOs. That is because emotions are not separate from intelligence, but integral to it – which is the reason that context based AI no longer seeks a Spock like program that “sees through” emotion. Let’s not go into the ethnography of emotions right now – that is a whole other chapter. The fact is that computers are very good at storing cases, segmenting case units according to some principle, surveying large numbers of cases, and establishing patterns. This is essential to representing context – which is not a matter of “logic” so much as a matter of structure. Emotion is great at structure. Realizing that the firm is a unit in which exchanges have to do with status seeking, emotional gratification or its delay, etc., is the necessary preliminary to replacing the CEO with the expert system.

There are a lot of researchers out there working on this. Yet, you read very few academic business profs writing about it. I wonder why? Could it be, uh, $$$$? The inflated status of the CEO was due to many things – the usual Marxist predicted decline of profit in the 70s, the new de-regulating atmosphere of the 80s, the success in overthrowing standards that had been built around the principle-agent problem, etc. But in order to gain public acceptance, business profs played an essential role in shilling for upper management, down to shilling for the absurd takeoff of upper management salaries. The justifications were byzantine, baroque, and resistant to reality. And the culture that this left behind, among economists and business profs, still remains with us, with the incentives really piling up for apologetic academic work – post facto justifications for enormous rent-seeking activities.
Thus, don’t expect IBM to put on-line some CEO Big Blue any time soon. But the theoretical ability to do so is already out there.