Daniel Susskind and why AI changes tasks before it replaces job titles

Does AI need to feel, judge, or think like us to deliver better business results than a human?

Economist Prof. Daniel Susskind has researched the future of work for years. My main takeaway from his latest talk: we overestimate how important human similarity is for automation.

Deep Blue did not beat Garry Kasparov through chess intuition. It used computing power, search, and systematic evaluation of variants. Watson won Jeopardy without knowing that it had won. It did not understand the questions like a human. It processed language, clues, and data differently.

Both examples are long before ChatGPT. That is exactly why they are useful. They show the point: what matters is not whether a machine copies the human path to a solution, but whether a task can be solved better through a different path.

With generative AI, this principle is now arriving in knowledge work. It no longer works only in a specialist game or quiz show. It works with the tools and core systems companies use to get work done: Office, code, CRM, ERP, ticketing systems.

There it writes texts, creates analyses, prepares decisions, and triggers process steps that humans previously handled.

AI still has limits: context limits, hallucinations, weak verifiability, and unresolved responsibility. Still, AI can already solve specific tasks better than a human, without having classic human capabilities.

For the future of work, that matters. Every job contains many individual tasks. A controller analyses numbers, checks variances, writes explanations, and is responsible for the interpretation.

Some of those tasks can be done faster or better with AI. Others remain human.

The relevant question is not which job titles AI replaces. It is which tasks change how quickly, and what that does to the job profiles around them.

Source: Prof. Dr. Daniel Susskind, "The Future of Work in the Age of AI" https://youtu.be/KfIY-01GyJs