Joe Toscano, the AI ethicist featured in Netflix's The Social Dilemma and author of Automating Humanity, joins Sunny Ray to explain how his warnings about AI became the foundation for his company, Service Stories. Toscano describes his years as a Google consultant, his work advising antitrust cases against Facebook and Google, and his core concern that tech companies profit from private user data much like reading someone's mail. He argues search itself has changed, users now type full sentences instead of keyword strings, and AI engines answer directly rather than returning a list of links. Service Stories responds by turning real work orders from service businesses like auto shops and plumbers into factual, story based content that mirrors the questions customers actually ask, helping local businesses get cited inside AI generated answers instead of buried in search results. Toscano explains why platforms like Reddit briefly became AI training goldmines, how query fan out multiplies the ways a business can be found, and why he still is not anti-AI, only against irresponsible, exploitative use of people's data and attention.
AI ethicist from The Social Dilemma explains why he now builds AI content systems, not against them, from real work orders.
What was it like being on Netflix's The Social Dilemma? Could you walk down the street without getting mobbed?
It was a documentary, not exactly glamorous, but when it came out I had people screaming at me from cars in LA and other big cities. It has over 500 million streams and is the most watched technology documentary ever. It took over my life for six or seven years.
You wrote Automating Humanity and appeared in The Social Dilemma making the case against where AI systems were heading. Then you built an AI agent. Can you walk me through that turn?
I was never anti-AI, I always argued for responsible, value driven AI, which is why I never got in trouble with Google or Facebook even as a Google consultant. I only spoke about what was already public. I wanted to change how data was used. What we're building now is a high value AI service, the future of work, so I don't see it as conflicting.
What exactly were you warning about, precisely, without flattening it into just AI is bad?
Job loss was one of my biggest concerns, along with respect for privacy. My parents were postal workers, and opening someone's mail is a federal offense. Making money off people's private messages and chats is no different than opening their mail and selling it. I believe companies should be responsible for how they sell and use data, with real purpose limitations.
You were a consultant with Google My Business. What does that teach you about local discovery that people outside it get wrong?
Part of Google's business is built to capture more and more business information, partly to serve customers but also to help Google organize the world's information cheaply. If you build your profile into their database, they don't need to crawl your website as much. But no matter how big they are, they can't force adoption. User adoption isn't guaranteed.
Search used to hand you a list, now it hands you an answer. What exactly changed for a plumber or auto shop?
The basis of SEO is still there, it's not dead, but the surface area expanded. Old queries were three to five keywords, now people type twenty to thirty word sentences like they're texting a friend. Five keywords alone create quadrillions of combinations, so with longer queries the number of ways a business can be found is exponentially larger.
Service Stories turns work orders into content. Why is that raw material better than what an agency would write?
An agency could run a ticket through ChatGPT, but our system brings content in organically from real jobs, so nothing is made up. Google increasingly wants verifiably provable information, not generic AI content. Since we build from actual work orders, we can prove the work was done and that our technicians are genuine experts, which is what Google wants at its core.
What earns a business a place inside an AI generated answer?
The highest factor is still how closely you semantically match the query. Sources like Wikipedia and Reddit got favored because they're vast, encyclopedic, and written in natural human language, which made them useful training data. Schema and technical SEO help but aren't the heavyweight factor. We skip guessing by using the real questions customers already asked your front desk.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview