Making the future. Some existing machinery required.
Last week, I attended Datacom's Chief AI Officer Summit. Notable for two reasons: (1) I got some considerable and favourable feedback on WorkLattice (Canberrans must be very smart), and (2) I had the opportunity to discuss with peers what the role of the Chief AI Officer should be. This article is about the latter. I can discuss the former at considerable length if invited.
We discussed the capabilities the role needs, which got me thinking about what we might be overlooking while we work out the job description. For those doing this job, or about to inherit it, what should they understand now that will seem obvious with hindsight?
We don’t do enough thought experiments
A separate executive role suggests this technology needs attention the existing roles can’t give it. What, then, should that person understand?
Which leads us to a thought experiment:
💡 What might we learn by appointing a couple of predecessors: a Chief Electricity Officer around 1890 and a Chief Blockchain Officer around 2018? We have the unfair advantage of hindsight. What might they have missed?
The Chief Electricity Officer, c. 1890
Recognise what else has to change before the technology can deliver. Our electricity officer would have needed patience with an unfinished world. Even in 1901, machine tools generally weren’t designed to connect directly to electric motors and having a motor didn’t mean having machinery ready to use it. The surrounding equipment, supply and engineering needed to catch up. The officer needed to distinguish a limitation of electricity from a limitation of the machinery using it and know what would have to change.
Even GE’s own factory still relied heavily on belts, pulleys and shafts in the late 1890s. The company literally making the future didn’t yet look like the future. That seems well worth remembering before promising the board that everything would be sorted out by the next quarterly meeting.
Help people discover uses they wouldn’t know to ask for. Utilities developed “load building” because connecting households that mainly used electricity for lighting wasn’t enough. They needed people to find more things to do with it. Asking customers what they wanted could only take them so far: people had spent their lives getting things done without electricity. Our officer would need to show them possibilities and a blank suggestion box wouldn’t settle the matter.
Anticipate demands from the people becoming more valuable, not just resistance from those feeling threatened. During Swedish electrification there were increasing strikes over better wages and conditions, especially in expanding sectors. More productive machinery could make stoppages more expensive, strengthening workers’ bargaining positions. Our officer would need to think about the people becoming more valuable, who might reasonably want that reflected in their wages.
The Chief Blockchain Officer, c. 2018
The blockchain officer, who was objectively less successful than his peer from 100 years prior, faced a particularly awkward possibility: the right answer might be to use something else. If a conventional database could do the job, appointing a Chief Blockchain Officer wouldn’t change that. They might need to explain why the organisation should spend less on blockchain, including after getting a platform working. Maersk closed TradeLens because the industry collaboration and commercial viability weren’t there. Our officer needed permission to stop.
They would also need to work out who lost from the proposed efficiency. An efficiency gain for the industry could mean a loss of income for the people expected to make it work. For example, if a shared ledger removes the need to reconcile records, a partner charging for that service loses revenue. You might struggle to get them to join, that person may understand the proposal better than you think. Our officer would need to know whose costs fell, whose income disappeared and whether the people expected to contribute had any reason to do so.
There was also the problem of what went into the system. A faulty sensor or a dishonest person could supply information which the blockchain then faithfully preserved. Our officer would still need to establish whether the sensor worked, who supplied the information and why anyone should believe them. Somewhere along the way, a useful guarantee against tampering had acquired rather grander ambitions about removing the need for trust.
What it means for the CAIO
That leaves the CAIO with the tension between changing too little and believing too much.
Electricity challenges timidity: familiar equipment and existing demand could obscure the opportunity.
Blockchain challenges enthusiasm: a working system could conceal some fairly unhelpful incentives and information nobody had verified.
The CAIO needs to question the assumptions inherited from the existing business, and those arriving with the new technology. Then work out what would actually have to be true for the promise to hold: which pieces are missing, who benefits, and who or what we still have to trust.
That’s first principles thinking, applied to both the business you have and the future you’re being sold. I appreciate that “think carefully about the problem” is an underwhelming conclusion after appointing two imaginary executives, but here we are.
I’m a systems designer by trade who got curious about AI. If you’re working out what AI should mean for your business, I’d be happy to help you think it through. Get in touch.