We’re about to find out what happens when you skip a generation of engineering development.
Last month Nolan Lovett published a rigorous paper about something I’ve been worried about for a while:
How do we train the next generation of great engineers in an AI world?
Nolan calls this “Validation Tether”: effective use of AI tools depends on deep domain expertise, but AI is eliminating the grad-level work through which that expertise forms in the first place.
If you can’t struggle through learning about something, you don’t achieve mastery. If you don’t achieve mastery, how can you sign off on a bridge that’s supposed to be there for 100 years?
The economics are what concern me. The cost of developing a good engineer falls to the organisation that develops them, but the benefits of a good engineer are shared by the industry at-large because they can move between companies.
It doesn’t take a genius to realise the incentive of a firm is primarily to make profit, often short-term profit, and that this scenario is going to quickly result in a race to the bottom, sacrificing development of the future generation for today’s profits.
The consequences of these sorts of decisions are not going to show up for 15-20 years. In the 2040s we might be scratching our heads, wondering where all the good engineers went, when it was what we did today that actually sealed our fate.
You can’t rely on a competitive market to solve a problem when the rational short-term move produces collective long-term damage. Engineers Australia and the bodies that govern our profession need to be putting guard rails in place to ensure we have the future we want.
Engineers Australia - batter up.