Peripatetic business wonk and pundit’s pundit Simon Wardley wrote recently that the AI bell tolls more for people in the executive suite than it does for those in more traditional positions.
From my experience, most CEOs tend to demonstrate poor situational awareness, an inability to decipher doctrine from context specific play and the boardrooms are more akin to alchemy, gut feel and whatever is popular in the HBR than to chess playing masters. Various studies have questioned the impacts of CEOs e.g. Markus Fitza’s study demonstrated that the CEO effect on firm performance varies little from chance.
But how might this unfold? After spending three days talking about the Analytics of Things in Hamburg, I have some ideas.
First: Changing the time horizon
We don’t have an Internet of Things; we have a lot of things, on the Internet. But that’s quickly changing. Big industrial companies like General Electric and Siemens are connecting all their devices, providing real-time data on the state of an organization.
Some of those algorithms live at the edge, in the machines, factories, mines, and motors, constantly tweaking and adjusting and optimizing. GE, for example, has sensors that constantly adjust the blade angle and orientation on wind farms, with each tower running micro-experiments, seeking more power, and telling its peers. The wind farm learns.
But much of this data, particularly exceptions to what’s normal or expected, find their way back to a central data repository. Once known as a data warehouse, today Big Data types term this a “data lake.” This data can’t possibly be processed by human hands; rather, it relies on algorithms to do so.
Algorithms scrub what goes into the lake, cleaning and validating it. Algorithms swim around the lake, looking for what might be useful or unusual. And algorithms query the data, trying to find ways to sort and organize it, often based on the content itself or guesses at what disparate kinds of data might be related.
All this means that information a business used to have monthly, or quarterly, it now has instantly.
Second: The new liability
When the company can know everything immediately, new kinds of liability emerge.
- Insider trading: Financial filings have to happen constantly. Insider trading is when someone with foreknowledge of matters that might affect stock price acts on those things before the general public can do so. The window of opportunity for insider trading is huge when insight happens in real time but filings happen quarterly.
- Negligence: Courts generally recognize negligence when someone didn’t exercise best practices based on what they knew at the time. Real-time, connected companies no longer have the excuse, “I didn’t know this was happening.” Whether it’s fraud, or discrimination in HR, or embezzlement, or antitrust violations, or myriad other white-collar crimes, as algorithms get better it will be harder for executives to use lack of knowledge as an excuse.
Third: Conditions of insurance
In 2014, in the U.S. alone, companies spent $2.9B to insure their directors and officers against “alledged wrongful acts.” The number of ways that officers can be held liable increases vastly when the enterprise is connected and running in real time, but its humans are disconnected and running at human speed.
Several insurance firms offer discounts when consumers are willing to share behavioral data. Progressive Insurance, for example, puts a sensor in cars that detects things like speed and sudden braking, and charges consumers less for the privilege.
I find the term “opt-in surveillance” disingenuous. A better way to say it is “paying a premium for not being surveilled.” In other words, because Progressive has to turn a profit, it will eventually make sensors the norm, and penalize those who aren’t willing to disclose their driving habits.
The same thing will happen in D&O Liability insurance: Insurers will insist that boards run liability-sensing algorithms to warn the executives of potential problems as soon as they occur, in order to demonstrate to courts that the business acted in the best possible fashion as soon as it became aware of the situation.
Fourth: From fear to greed and power
A smart, cynical salesperson I once knew told me people do things for four reasons: They want to get laid, paid, made, or unafraid. The initial motivation for bringing this kind of company-wide algorithm is the fear of punishment or retaliation when someone is wronged. But given what such an algorithm might do, it will quickly appeal to other motivations.
An executive team armed with such information will be able to adapt and react to things better, and hopefully make wiser decisions. It will also be able to control its organization better than others, for better or for worse. And so soon, what began as protection will become advantage; defense will turn into offense.
Fifth: The machine board member
By this point, algorithms will have a seat at the boardroom table. Executives will increasingly check with those algorithms, which will handle much of the productivity-centric busy-work. Kevin Kelly talks about this in his new book, The Inevitable:
Generally any task that can be measured by the metrics of productivity — output per hour — is a task we want automation to do. In short, productivity is for robots. Humans excel at wasting time, experimenting, playing, creating, and exploring. None of these fare well under the scrutiny of productivity. That is why science and art are so hard to fund. But they are also the foundation of long-term growth.
Algorithms, borne of fear over liability from real-time information, are a good Trojan Horse for the “sentient enterprise” — another buzzword that’s used to describe a connected, semi-autonomous business. That’s because they’ll soon morph into advantage and power, possibly with troublesome effects.
Cory Doctorow has said that science fiction tells us a lot about the world of today, and that scifi films about AI and robot overlords are actually about nameless, faceless corporations. Surely a race to an algorithmically governed business is one of these fears?