The end of the road for “experts”?
As of 2023, companies hire job titles such as “full stack” web developer, mobile application developer, data scientist, UI/UX (user interface, user experience) designer, database designer, DevOps engineer, business analyst, and a similar array of managers for these functions. People spend years becoming experts in these areas. I can still remember how insulted I felt as a young server-side developer the first time a manager asked me to do a client-side development task
Let’s call the application of previously-learned knowledge and skills routine expertise.
I hope I am not the first to tell you that machines are rapidly gaining the ability to apply routine expertise. Routine expertise will be (or has already been) automated by generative AI and Language Learning Models. Job titles like the ones I listed above will be obsolete in 3-5 years. They will continue to exist in uncompetitive organizations.
What can’t machines do?
Training ChatGPT is energy and computationally intensive, plus the human effort required. For now, humans are better at learning and adapting than machines are.
A human may be more able to defy convention. For instance, while a language model might suggest UI/UX designs based on historical data, a human might invent a novel solution that breaks conventions.
For that matter, conventional wisdom is sometimes wrong or limiting, requiring the insight of a nonconformist human. And the trainers themselves have biases. If ChatGPT had been trained on the prevailing orthodoxy of Copernicus’s day, it would have said that the Earth was the center of the universe.
We’ll call all this learning expertise to distinguish it from routine expertise.
What does GLAD stand for?
Craig Larman has proposed the acronym GLAD to stand for Generative AI LLM (Language Learning Model) Assisted Development. (I guess GAILLMAD contains more capital letters than he can stand.) Craig describes himself as a junior GLAD developer.
What will work look like in the future ?
From The New New Product Development Game by Takeuchi and Nonaka (1986):
Under the rugby approach, the product development process emerges from the constant interaction of a hand-picked, multidisciplinary team whose members work together from start to finish.
With a focus on learning expertise, a small team of product developers will move from problem to problem much faster than your lumbering overstaffed departments do it today. In LeSS we have always emphasized multi-learning to eliminate queuing and handoffs. Soon the economics of this will be unavoidable. A team that’s serious about learning will also try mob programming.
Organizational designs (structures and policies) will have to be simplified to adapt to more complex work.
Will these super-teams do 15 minute daily meetings and estimate work in Fibonacci numbers?
I hope not! But the spirit of Scrum described by Takeuchi and Nonaka will live on.
Please view Craig Larman’s lecture on this topic below.