The Terminator myth, where an army of robots descends on the workplace to displace human beings, is a misconception. While machines do displace humans from particular tasks, they also complement them in other ways. This complementarity takes two forms:

1. Direct Complementarity: Machines can make human workers more productive and efficient at certain tasks. For example, a taxi driver can use a satnav system to navigate unfamiliar roads, or an architect can use computer-assisted design software to design larger, more complicated buildings.

2. Indirect Complementarity: Technological progress can expand the economic pie, creating new industries and tasks that require human labor. As productivity increases, incomes rise and demand grows, leading to the creation of new roles and opportunities for displaced workers.

The key point is that technological progress does not simply substitute human labor; it also creates new ways for humans to contribute and thrive. The threat of technological unemployment is real, but it is a "good problem to have" as it reflects the success of making the economic pie larger.

The second myth that the speaker addresses is the "intelligence myth" - the belief that machines have to copy the way human beings think and reason in order to outperform them. This view was prevalent among economists who thought certain tasks, like driving a car or making a medical diagnosis, couldn't be readily automated because they required creativity, judgment, and intuition that were difficult to articulate.

However, the speaker argues that this view is becoming increasingly outdated. Advances in processing power, data storage, and algorithm design mean that the routine-nonroutine distinction is diminishing. Machines can now perform tasks in very different ways from humans, without needing to replicate human intelligence.

For example, the system that can diagnose skin cancer as accurately as dermatologists doesn't try to copy human judgment or intuition. Instead, it runs a pattern recognition algorithm through a vast database of past cases to identify similarities. The fact that human doctors can't fully explain their diagnostic process doesn't limit the machine's ability to outperform them.

Similarly, IBM's Watson computer was able to beat human champions on the quiz show Jeopardy, not by replicating human reasoning, but through a very different approach. The speaker argues that our limited understanding of human intelligence is far less of a constraint on automation than it was in the past. As machines perform tasks in novel ways, there's no reason to think human capabilities represent any kind of ceiling on what machines might achieve in the future.

The third myth that the speaker addresses is the "superiority myth." This myth suggests that as the "lump of work" grows and changes due to technological progress, human beings will necessarily be best placed to perform the new tasks that emerge.

However, the speaker argues that this is a fallacy. While it is true that technological progress expands the overall amount of work to be done, it does not necessarily mean that humans will be the ones to perform these new tasks. As machines become more capable, they may end up complementing and enhancing each other, rather than complementing human workers.

For example, in the case of driving cars, the speaker notes that while GPS systems currently complement human drivers, in the future, self-driving cars powered by software may simply make the machines themselves more efficient, rather than enhancing human performance. Similarly, as the economic pie grows and changes, the new demand may be better met by machines rather than human labor.

In essence, the speaker argues that "demand for tasks isn't demand for human labor." As machines become more advanced, they may be better positioned to take on the new work that emerges, weakening the helpful "complementarities" that have historically favored human workers. This, combined with the increasing substitution of machines for human labor, paints a troubling picture of the future of work.

The future of work is both troubling and exciting. The threat of technological unemployment is real, as machines continue to encroach on tasks performed by human beings. However, this is a symptom of our success in solving the traditional economic problem of making the economic pie large enough for everyone to live on.

While the balance between machine substitution and machine complementarity has historically favored human beings, this balance is shifting. Advances in artificial intelligence and robotics are strengthening the force of machine substitution, while weakening the helpful complementarities that have benefited workers in the past.

Solving the challenge of ensuring that everyone can enjoy the material prosperity generated by our economic system, in a world with less work or even without work, will require us to think in very different ways. There will be much debate about the appropriate solutions, such as various forms of universal basic income. However, this is a far better problem to have than the one that haunted our ancestors for centuries: how to make the pie big enough in the first place.