This is how Ai Will Change The World


Imagine asking a human to fetch you a cup of coffee. You don’t mean for this to become their sole purpose in life, leading them to extreme measures like harming others just to get that coffee.

Instead, a human would consider all other factors and understand the broader context of the request.

However, the way we currently build AI systems is different. We give them a fixed objective, and the algorithms require us to specify everything in the objective.

For example, if we tasked an AI with fixing ocean acidification, it might do so efficiently but at the cost of consuming a quarter of the Earth’s oxygen. This would lead to catastrophic consequences for humanity.

Humans inherently know they don’t have all the answers and will ask for clarification when needed. AI systems, as they stand now, do not have this capability.

They operate under the assumption that they fully understand their objective, leading to potential disastrous outcomes if not carefully managed.

Building Smarter AI

To avoid these issues, we should build AI systems that recognize they don’t know the full objective.

These systems would exhibit behaviors like asking for permission before taking drastic actions, ensuring control over the AI comes from its uncertainty about the true objective.

The Impact of AI on Employment

The concept of “technological unemployment,” a term popularized by John Maynard Keynes in 1930. The idea is simple: as machines take over tasks, humans will lose jobs.

This is already evident in e-commerce warehouses where robots retrieve shelving units, but humans still pick items from the shelves because it’s too complex for robots to handle—at least for now.

As AI continues to advance, the potential for robots to perform these complex tasks could eliminate millions of jobs.

This brings to mind E.M. Forster’s story where civilization becomes entirely machine-dependent, leading to a loss of knowledge and understanding.

This scenario is echoed in the modern film “WALL-E,” where humans become helpless due to their reliance on machines.

Preparing for General Purpose AI

The arrival of general purpose AI won’t happen overnight. It’s not an all-or-nothing event; rather, its impact will gradually increase.

While some experts predict we will have general purpose AI by the end of this century, some are more conservative, believing it might take longer due to the complexity of the problem.

Check out this video on the topic:


Stuart Russell’s insights offer a fascinating glimpse into the future of AI, highlighting both its potential and the challenges we must address.

As AI technology advances, it’s crucial to develop systems that can navigate uncertainties and ensure they align with our broader human values.

Only then can we harness AI’s power without compromising our safety and well-being.