Jihad and Human Inputs
More rational than you think….
The first thing to understand is that technology without human psychology creates, at best, wildly mismanaged efforts. At worst it’s downright dangerous. So my first thought when Dario Amodei announced that 5% of Anthropic stock would go toward a “public benefit fund” was that it sounded like a cross between Trump’s plan to pay every one $5,000 to vote Republican and George Costanza’s Human Fund. Something performative and vague to get through a IPO, a mid term or, in Costanza’s case, a whacky Festivus season. Yet, Amodei isn’t on the short-list of post-AI overlords because he’s a moron - he knows that the push back is only just starting.
Before getting into questions like AI destroying our jobs and then an unemployable humanity, we might want to ask how well your optimized AI strategy will even function when applied to a universe of messy humans. History has some vivid examples: Socialism – that idea that will neither fly nor die – is a data driven, centrally planned model operating on system efficiencies for the greatest utility for the common good. Silicon Valley’s latest iteration is called “effective altruism.” Either way, it hews to a certain superficial logic while failing in very predictable ways when it touches human psychology or reality.
For a more effective way to harness human buy-in, you could do worse than look to something as superficially irrational as jihad. As I wrote about in Pothole of the Gods: Holy Wars, Proxy Wars and Bad Ideas it works on a human level and has since close to the beginning of civilization. While the outputs of holy war are crazy, the human inputs are often hyper-rational. I’ve got some back-up on this. In a 2009 paper, Why Are There So Many Engineers Among Islamic Radicals?, sociologists Diego Gambetta and Steffen Hertog found that engineers were disproportionately represented among the individuals they studied. In Egypt, the nickname for the Muslim Brotherhood is the “Engineer’s Union.”
Which brings us to the uncomfortable conclusion that radical terrorists, AI and your finance department are all running on similar and unambiguous models in an attempt bring order and efficiency to an untidy world. And only one of the above is really, really good at motivating people.
Here are three reasons why:
Humans Hate Uncertainty:
The world is a hairy and unpredictable place. Lacking a complete picture or the super-intelligence to process the information if we did, people developed survival calculations that heavy on pattern recognition and context, and light on actual data. One explanation for the strange case of hyper-rational engineer sorts getting radicalized is that they have a mindset attracted defining problems, establishing constraints and finding solutions. These are useful habits when working with bridges, electrical circuits or industrial processes, but less so when applied to societies full of competing interests, conflicting values and people reacting to personal pain-points.
A major technological change does more than alter how work gets done. It alters the context of the people the change affects. It changes what employees can expect from the organization. Will their expertise retain its value? Will the skills that keep them employed become obsolete? Will management use increased productivity to grow the business or simply reduce the headcount? The finance department sees an efficiency gain. Employees see a threat to their livelihoods and status. Both are responding rationally to different risks.
People Need Something to Believe:
If uncertainty is the problem, belief is one way humans manage it. We create frameworks of belief and assumptions that bring order to the world. These models are self-serving but useful and a little self-delusion, but that is a small price to pay for a model of the world that gives us a comforting level of certainty needed to focus on the wide-context details.
A expanding into a new market, rebuilding its technology or retraining its workforce may have a perfectly rational financial case for the firm, but people don’t think in spreadsheets. For that matter, most of us don’t think in math. People need a reason to pursue an objective, not merely an instruction to hit a target. Common objective and culture enables common strategy, but that requires a credible account of the future, tangible outcomes and why they matter.
People Are Tribal:
And the stronger the common culture, the better. We are social animals. We look to our peers for information, approval and protection. Belonging is a powerful force for cooperation with shared symbols, common experiences and a clear distinction between insiders and outsiders. Disrupting that cohesion is tricky.
Management labels such resistance a failure to understand the technology or new corporate goals. Sometimes the real issue is that the change threatens people's place in the organization. Treat adoption as a social process, not simply a technical deployment.
THE FIELD GUIDE
While wide-context human solutions are heavily policed by quantifiable narrow-context logic, the opposite isn’t true. The end results are spreadsheet ready solutions that increasingly don’t map onto human psychology. A technology optimizes a process, but only if human inputs continue to work. The more humans are replaced, the more a system move from human reality - resulting in, say, call centers that function like terrorist cells.
People seek security in an uncertain world. They need a purpose that makes effort worthwhile. And they care about the groups in which they work, compete and build their identities.
Uncertainty: What are people afraid of losing? Identify the risks any change creates for employees, customers and partners. Explain how unresolved questions will be handled. Watch for a gap between the risk leadership believes it is managing and the risk everyone else thinks they are facing.
Belief: Why is this worth doing? State the intended outcome in terms people can understand and connect to their work. Explain what success will change for customers and the business. If the only persuasive argument is that the shareholder will be jazzed, the case is not finished.
Tribe: Whose cooperation matters? Identify the trusted people, informal networks and professional identities that influence behavior. Bring those people into the process before decisions become irreversible. Make sure employees have a credible way to gain in the new system rather than simply lose it in the old one.
Then run one final check: Are you measuring the result you want, or merely the behavior that is easiest to count?
