My Most Recent Podcast
My most recent guest on Brave New World was Paul Shapiro, CEO and co-founder of The Better Meat Co. Paul is also author of the book Clean Meat, which chronicles the growth of the alternative meat industry and the people who shaped it. Paul is doing some very cool things, like brewing proteins to make foods that replicate or enhance the meat experience, and to create new kinds of nutritious foods. I can imagine that the next generation of such foods will use gene editing technology.
A side effect of this innovation is that it will spare real animals from mass slaughter, and reduce water pollution, deforestation, and greenhouse gas emissions. And perhaps we’ll have fewer health problems that result from all that animal fat that ends up in our arteries. A trifecta.
So, tune into Brave New World to listen to what Paul Shapiro has to say.
Tech on the West Coast
I’ve been traveling in the bay area with a group of bright and colorful Stern Tech MBA students, teaching a class on Tech Innovation. Many thanks to Sarah Ryan and Alphonsina Frias Vargas of Stern for organizing visits to the big tech companies and making the class hum. As luck would have it, Alfonsina’s partner, Tim Smith, was playing drums for rhythm and blues singer Emily King at the Herbst Theater in San Francisco last weekend, so I tagged along to an amazing show.
I’ve told Sarah that she should run for public office, and aim for nothing less than the top job. So, when she does become president of the United States, I’ll be able to say that I saw it coming way before anyone else. Go Sarah!
Between the visits to companies, we’ve had stimulating discussions about Artificial Intelligence, which is front and center. I’ve grown up with AI over my career, but the recent advances we are seeing are nothing short of astounding. Every visit to a tech company and discussion with the students has featured ChatGPT3, the chatbot that has taken the world by storm in a few short weeks.
My view is that ChatGPT3 has pushed Artificial Intelligence from being an application towards becoming a “general purpose technology,” or a GPT, not to be confused with the language model GPT3. As I wrote in my last newsletter, ChatGPT3 seems to “understand” us. Whether it really understands anything or appears to understand is a deeper question, but as philosopher Dave Chalmers asks, if you can’t tell the difference, does it really matter? More on that below.
The larger point is that AI has fundamentally altered our relationship with machines, which are now able to engage with us on our terms, not theirs. Until now, it has been difficult to communicate with them. We’ve had to do it through keyboards, mice, or writing programs. Conversational AI changes everything. Every single interface will be impacted. Just like electricity, which changed every factory, AI will change every single interface with machines. It’s nothing short of a revolution.
The Next Tech War
ChatGPT3 is the beginning of a new war between Google and Microsoft. Google, the seemingly unassailable tech titan, is suddenly highly vulnerable. Its amazing search engine looks primitive compared to ChatGPT3.
Ten years ago, I published an article in Wired Magazine, with the title, “Google in Jeopardy,” encouraging IBM to take on Google using its Watson system. Watson had just beaten the human Jeopardy champion, which was an impressive feat that required sophisticated natural language processing. I suggested making Watson conversational, compared to Google’s search engine, which merely returned results. Why not let Watson converse with people in natural language, I suggested, about things like health, real estate etc., and resort to Microsoft’s Bing search engine when it couldn’t advance the dialog? Microsoft and Yahoo had lost the search war to Google.
IBM’s leadership invited me to meet with its scientists and leadership and heard me out. But they said they were a “Business to Business” (B2B) company, not a “Business to Consumer” (B2C) company. Now, there’s an example of a lack of vision and courage. Amazon was also a B2C company, but Jeff Bezos didn’t see why it had to be that way. The rest is history. Watson is practically in the graveyard now, a wasted asset.
Maybe IBM was right, not about its B2B excuse, but because it didn’t have enough AI horsepower to build on, and would look foolish by taking on the then invincible Google, the world’s first real AI company. IBM had survived a near death experience itself due to decades of poor leadership, and reinvented itself as a tech services company under Lou Gerstner, and things were going okay. Why take a reputational risk?
It’s a whole new game with ChatGPT3 that is backed by Microsoft, which has very deep pockets and a visionary leader. You can’t underestimate Satya Nadella. Like IBM, Microsoft had also fumbled, but under Satya, it’s a whole new tech company. And as of last week, Microsoft now owns half of OpenAI, the creator ChatGPT3.
Microsoft’s strategy is nothing short of brilliant. Both it and Google are serious companies, that can’t be releasing frivolous products. Google has its own language model, Lambda, which one of its engineers, Blake Lemoine, dared to suggest was sentient. Lemoine was put on leave. It’s the kind of action a staid company like IBM would have made. I hope Google isn’t becoming the new IBM. I’m a Google shareholder.
But ChatGPT3 is just fun. It’s more of a toy than a real product. Not yet anyway. Microsoft learned its lesson by releasing its chatbot Tay, which turned into a racist machine soon after its release, so Microsoft quickly killed it. Oops, sorry, it said. Even though it’s a toy, it is gathering staggering amounts of useful data from us, which will turn it into a real product in the not too distant future.
With ChatGPT3 belonging to OpenAI, Microsoft can afford to be silly and playful until it is a real product. Everyone is having fun with ChatGPT3. Critics are quick to dismiss it as something that doesn’t really understand anything, and hence, not worthy of being considered as having “general intelligence,” which is the holy grail of AI. it’s just a gimmick, they argue, and won’t be taken seriously.
They couldn’t be more wrong. In a TEDx talk I gave five years ago titled “When Do We Trust Machines,” I argued that we trust machines when the costs of error are low. I could list numerous low-risk applications of GPT3, that include proofreading and summarizing documents, repurposing existing content towards new purposes, and creating new content – all of which are expensive to do at the moment. ChatGPT3 could do this in the blink of an eye. It may not understand the world the way humans do, but that’s irrelevant.
The history of tech innovation is littered with cadavers of dominant companies that didn’t take real threats seriously until it was too late. I don’t think Google falls into that category. It sees the threat from Microsoft. The question is, how will it respond?
Hi Vasant, I agree that there are some immediate applications of systems like ChatGPT. One I discovered is researching somewhat arcane technical problems which normally could require reading a number of documents. An example. What software necessary to install the slurm database daemon on slurm? In seconds , ChatGPT responded with appropriate step by step instructions. It performed equally well on several other technical questions, seemingly summarizing results. I assume that Google's system, LaMDA will perform equally well, when the release it to the public. I believe that the next few years will be a continuing struggle between MSFT (OpenAI) and GOOG (LaMDA) to see who will become the dominant player in Search 3.0. MSFT struck the first blow, but GOOG has scale and data and very good people. Stay tuned.