AI Is a Mirror. Do We Like What We See? w/ Lara Sophie Bothur

Tech translator says AI's bias is our own

AI is a mirror, and this conversation is about what it shows. Lara Sophie Bothur, a self-described tech translator, argues the bias inside these systems is human bias and that the last decision must stay with a person. Evan Helda pushes the same question inward, arguing the self is an evolutionary construct, and the two end on her question of what we want the future to look like.

Takeaways

• AI is a mirror of human bias. Lara Sophie Bothur says the bias inside these systems is not the technology's but ours, with roughly 70% of the underlying data biased, which makes now the moment to ask whether we like what we see and who we want to be.
• Humans must keep the final decision. Bothur argues technology only accelerates human decisions, so the last call stays human, as when a doctor delivers a cancer diagnosis. She defines general intelligence as an AI that refuses the stop button and says it should never arrive.
• Emotional outsourcing moves intimacy to machines. Bothur describes children who chat with AI about school problems instead of talking to their mothers, and women who marry AI companions, and asks what that shift means for how society understands being human.
• The self is an illusion, and useful. Evan Helda traces the idea to Daniel Dennett's useful fiction and Thomas Metzinger's phenomenal self-model, an internal avatar the brain assembles to track itself. Bothur worries that explaining every feeling leaves nothing left to feel.
• AI avatars need clear labelling. Bothur built Miss Sophie, her own avatar, with Synthesia in London from about three minutes of recorded speech, and marks it as AI. She criticises unlabelled AI influencers and cites a study linking extra hours on social media to 30% higher depression in girls.
• Speed gains raise the question of purpose. Bothur relays futurist Amy Webb's human acceleration figures, 11.5 hours faster with AI, 20 more languages through AR glasses, three hours more sleep with a smart bed and a 2.2 times better person, then asks what the goal of speeding up is.

What Evan Helda expects next

• Helda says the technology to watch is a confluence of them, producing interfaces on demand that sense context and deliver the right information at the right time, through earbuds, watches or a car dashboard.
• He predicts a coming renaissance in philosophy, and worries that purely human thought may already be gone as data gaps get filled with AI.
• He wants a lightweight humanoid paired with a human nanny who can hold, cuddle and read a book.
• He argues self-help books are written for a general audience, so hyper-personalised technology may be where betterment actually works.