How Durable are Diamonds?

Audio Narration

0:00

/452.31020408163266

1×

Most of you would say diamonds are pretty durable. If you’re a chemist, a  materials scientist, a toolmaker or a jeweller, you might say something a bit more precise/geeky like ‘They top the Mohs scale (10/10), meaning nothing can scratch a diamond except another diamond. But they are brittle, and a well judged blow along a crystal plane can cleave them asunder’. If you’re a socio-economist specialising in corporate structures, you might say ‘Diamond structures are potentially durable if carefully managed. But if they emerge rapidly through AI job displacement, they can be the first stage of an unstable inverted pyramid’. 

I’ll unpack this a bit more  in a moment, but here are a couple of posts I read last week and a perspective from an Uber driver.

1. The excellent, and often amusing, Bloomberg columnist Matt Levine has almost as much to say these days about AI as he does about investment banking (He previously worked for Goldman Sachs). So he was absolutely in his element    posting last week about Open AI’s latest move to disrupt investment banking. As Levine observes, junior investment bankers are well paid, but they do work 110 hours a week, largely formatting spreadsheets, aligning the $ signs correctly and trying not to leave the wrong corporate logo in copy/pasted client presentations. So $150 an hour doing exactly the same at Open AI with zero stress must be an appealing prospect. Open AI are of course on a mission to train an agent that will do everything that a junior investment banker does, but 10 times quicker, 10 times cheaper and with 10 times fewer mistakes. If this works out as expected, the need for investment banks to employ small armies of junior bankers will evaporate, although middle rank and senior bankers will still be needed to martial the AI foot soldiers, pick up restaurant bills, play golf and more generally close the big deals. So the base of the organisational pyramid in the investment bank is eroded, leaving a more cost effective structure in the short term but serious succession questions in the longer term; where will the next generation of senior bankers come from   if the bottom of the pyramid shrinks too much?
2. Last week’s weekly Applied AI newsletter by Aaron Holmes  picked up on Andrej Karpathy’s recent podcast with Dwarkesh Patel, in which Karpathy, a founding member of OpenAI, opined that AGI is a decade away – very possibly true, depending on how you define Artificial General Intelligence. Holmes “Reality Check On Agents” highlights the need for agent supervision, including this quote from Sahil Lavingia, an eCommerce CEO: “I think it’s interesting that I can replace some $400,000-a-year jobs [with agents], but I can’t completely replace them. I just have someone else I pay $400,000 to manage agents to become four times as productive and then fire three people."
   Personally I find articles like this one, playing down the significance of AI agents, quite confusing. Of course agents needs supervision, just like junior programmers, junior bankers, junior architects and junior doctors need supervision. But even if agents need a bit more supervision than some junior humans, surely the point is that the business needs fewer junior humans. Economically attractive in the short term, but once again this highlights the succession dilemma; where will the next generation of senior professionals come from if the base of the pyramid shrinks too much?
3. My regular conversations with Uber drivers, hotel staff and fellow sufferers on Network Rail often counterbalance the pervasive AI optimism amongst my tech industry colleagues. Last week, an anecdote from an Uber driver was particularly stark, with his conviction that AI is already ruining humanity. He attributed the 50% drop out rate on some architecture degree courses entirely to students worrying that all junior architecture jobs will have been swallowed up by the AI monster before they qualify’. There is much going on here, but it’s interesting that his own son dropped out and is now driving a bus; a job that ironically may be displaced by AI before architecture is displaced, but the risk of wasted time qualifying as a bus driver is so much smaller than the minimum of 7 years to qualify as an architect. Careers with extremely long training periods are particularly exposed to this legitimate concern by students investing the best years of their lives in a really hard training grind. Once again, the base of the pyramid in architecture may become too small to support the apex.

Now finally back to the question of a diamond structure. If the requirement for juniors diminishes year by year  as AI becomes more capable, a profession will naturally adopt a diamond shape – narrow at the senior apex, fat in the middle with mid-level professionals still climbing the ‘Up or Out’ ladder, but narrow at the base due to the sharply diminished intake. As the years progress there are two possibilities. If the profession or company takes no action to change an ‘Up or Out’ culture the diamond will become columnar as the base becomes the middle and may even become an inverted pyramid which is intrinsically unstable and will topple. Alternatively, an organisation can in principle change its culture in two substantial ways to make the diamond evergreen. Firstly career expectations and opportunities must accommodate career journeymen – i.e. a stable cohort of employees who get long term job satisfaction and good remuneration from a stable role in the central band of the diamond. Secondly, the organisation must properly fund and encourage the right sized intake to replace natural attrition in the central band. Junior recruitment numbers cannot be dictated by short term economic comparison with AI agents , but must be managed and funded with a long term view of sustaining the organisational fabric.

There are professions and trades which already encompass this journeyman mindset – nurses, teachers, electricians, engineers and many more. These structures do have the huge additional benefit that returning to the centre after a career break is much more straightforward than the ‘Up or Out’ ladder. Finishing on a positive note, the reduced need for highly paid junior bankers and management consultants will make significantly more very smart and very hard working candidates available for the most impactful career path that will always exist – entrepreneurship; I.e. disrupting the status quo, driving innovation and increased productivity.