Artificial general intelligence is risky by default

Should we worry about AI?

Connor Leahy is an AI researcher at EleutherAI, a grass-roots collection of open-source AI researchers. Their current ambitious project is GPT-Neo, where they’re replicating currently closed-access GPT-3 to make it available to everyone.

Connor is deeply interested in the dangers posed by AI systems that don’t share human values and goals. I talked to Connor about AI misalignment and why it poses a potential existential risk for humanity.

All sophisticated solutions start extremely simple

There’s always a temptation to launch a fully built product with more features and capabilities than existing competitors. It’s exciting to build the next Google, the next iPhone or the next SpaceX, isn’t it?

This temptation is dangerous because even the most successful products in a market had simple beginnings. No product arrives in the market fully fleshed out. The company behind a successful product has developed its internal capabilities and know-how about various tiny but important details over a long period of time. On day 1, a startup simply cannot match such capabilities.  ...  Read the entire post →

Is the world becoming better?

This essay is inspired by the book Factfulness where the key idea explored is that the world has witnessed significant progress over the last few decades, but most people are unaware of that fact because they hold distorted views.

For example, most people overestimate the percentage of the human population living in extreme poverty (which is defined as having an income of less than $2/mo). It's 10% (as of 2017) but even the most educated people get it wrong.

Reality is an evolved illusion

Do we see reality as it is?

I discuss this question with Donald Hoffman who is professor emeritus at the University of California, Irvine. He studies consciousness and perception from an evolutionary point of view. His research has led him to make a bold claim that while we do not yet know what the underlying reality could be like. Rather, reality as we know it now - including space, time, and objects - is a useful fiction that evolution invented for us.

People don’t like using technology

There’s an inherent tension between how an engineer sees the product under development and how a potential user sees it. Engineers get excited by the technological advances and configurability of the gadget. A user sees all such complexity as overwhelming and off-putting. It's easy for a creator to forget that the user has a life to live and using their product isn’t a highlight of her life. Rather, it’s likely a chore.

Most surgeries are ineffective

Do surgeries work? Most of us assume they do, but is there any scientific evidence that they do?

In this episode, I talk to Dr Ian Harris who is a Professor of Orthopaedic Surgery at the University of New South Wales in Australia. He is a practicing orthopedic surgeon specializing in trauma surgery. Outside his practice, his research interests broadly cover the topic of surgical effectiveness and clinical research.