Food Tribes & AI Echo Chambers
The future of Generative AI will be defined by customizable AIs that can amplify the entire spectrum of opinion about food, nutrition, or any other topic that their creators desire.
Before the Internet, most of Western society was beholden to just a handful of powerful media outlets that dictated the mainstream narratives in almost all parts of life. Dissent was there, it just hadn’t proliferated yet into the wild mosaic of opinions you can find online today. The democratization of creating and spreading new ideas far and wide via the Internet changed the world in ways that we are still processing.
The effects of the information age manifested itself in food by splintering eaters into a long series of food tribes, each bound by a unique set of values, preferences, and buying behaviors outside of the mainstream. From keto to veganism to gluten-free diets to biohackers, food tribes run the gamut. Pre-Internet, if you were a yoga-loving vegan living in the middle of nowhere, you were probably alone as far as your lifestyle choices were concerned. Today, just a few quick searches on any major social media platform will connect you with thousands if not millions of others who eat the same way as you.
This newfound ability for food tribes to connect undermined the core thesis of Big Food, which was to make a single product that would achieve mass appeal. People didn’t want to eat the big brands anymore, they wanted stuff that was made for them and their tribe. For the most part, this is a good thing, because we have more choices than ever and can all get what we want.
But the long tail of food also contains a lot of products that are essentially snake oil. Propped up by lifestyle influencers ranging from Goop to Liver King, and a word salad of claims that “have not been evaluated by the Food and Drug Administration,” we’ve reached a point in food where merely believing that a food is good for you is often as good as it actually being better for you.
So what does this have to do with AI? Well, the information we consume helps dictate the food we eat and Generative AI is great at creating information that can be tailored to represent any worldview. Like the food industry, where mass-market products dominated then ceded control to a ton of niche products, the same thing is happening in Generative AI, albeit at warp speed compared to the food industry.
While most mainstream internet users have been fixated with ChatGPT, there’s been a large, vibrant community of open source enthusiasts and developers tinkering with off-the-beaten path versions of chatbots and image generators that aren’t getting the same amount of press. I don’t think the average person understands how easy it is to download some code and create your own personal chatbot or image generator that runs locally on your computer and can be customized to your liking, with no content restrictions.
The current food revolution has been defined by upstart entrepreneurs unhappy with mainstream products, who corralled the resources to develop the products that they wanted to see on the market. It was and still is a difficult, arduous journey to creating your own food product, but enough people succeeded that we now have a wide variety of voices in the food industry. Open source Generative AI tools are enabling a similar trend on a much larger scale, as anyone who doesn’t like the mainstream AI tools on the market can easily create their own over a weekend with a beefy laptop.
Much like how Goop became an alternative source of health, food, and lifestyle information for those looking for something outside of the mainstream, there’s nothing to stop someone from creating a food focused ChatGPT alternative trained on any corpus of knowledge the creator desired. It’s becoming very easy to take every word that Goop or Liver King has written or spoken on the internet and use it to fine-tune a chatbot so it parrots their worldview. Every faction of the food industry, from non-GMO advocates to Monsanto and everyone in between, can create chatbots that will tirelessly and consistently propagate their opinions to the world.
I am a firm believer that the future of Generative AI will be defined by the ability for users to customize them to their exact specifications. The open source community has known this for a while and OpenAI has even gotten in on the action with their recent release of customizable GPTs, where users can customize their own chatbots using only natural language prompts. No coding necessary, which means if you know what you want and you know how to type, you can make your own custom chatbot in a matter of minutes. OpenAI will enforce guardrails to prevent custom GPTs from producing lewd, harmful, and false content, but as we’ve seen in the food and nutrition space, there’s a lot you can say about your product that’s not overtly false, but isn’t really true either. Between truth and lies is a wide canyon of food claims that are either unproven or supported with shoddy scientific evidence.
The user experience is markedly different when speaking with a chatbot than it is to conduct a Google search. With Google searches, at least you had pages and pages of search results where you could pick and choose the information sources you found most credible. A page of search results doesn’t offer up any opinions, it simply provides you with a menu of unadulterated webpages to choose from. But if you believe in the very plausible reality that chatbots may eventually replace search engines as our primary sources of information gathering, the chatbot you choose can affect your reality in profound ways. ChatGPT will soon have a foil in Grok, the “anti-woke” chatbot that Elon Musk is launching within X, so the idea that there will eventually be a chatbot that represents the entire spectrum of opinion is very real.
We already have giant factions of people who have built their own echo chambers, only following Fox News or MSNBC and filtering out any opposition media that doesn’t jive with their values. People tend to only want to hear what they want to hear because disagreement can be uncomfortable, yet necessary. What makes us think that we as a society will be suddenly more evolved and balanced in the presence of Generative AI that we won’t use chatbots to go deeper into those echo chambers?
Whatever the next iteration of Goop or Liver King fanatics look like, it’s not hard to imagine them only conferring with GoopGPT or LiverGPT for food and nutrition advice. No matter how hard the operators of mainstream chatbots like ChatGPT work to ensure all answers come out fair and balanced, anyone will still be able to build alternative chatbots on a different set of facts and have the same amount of reach as a multibillion dollar company like OpenAI has.
The internet was the first step in amplifying people’s abilities to express non-standard views to a wide audience. Generative AI and the legion of niche chatbots we are likely to see over the coming years doubles down on that ability. But humans still have agency in choosing what tools and information sources we’ll ingest. Chatbots make it temptingly easy to get a single answer about something without having to wade through pages of search results. This will encourage lazy online research for sure, but chatbots can just as easily give you a well-rounded, fact-based answer if you ask it to do so. Generative AI is a powerful idea amplifier, but it’s humans that guide them toward good or evil objectives.
When it comes to food, it’s up to us, not a chatbot or search engine, how to handle the new deluge of content that Generative AI will help create. We eaters will collectively decide if food tribes metastasize into deep fanaticism that foments snake oil, or if we will reject junk science and misinformation in search of truth. The internet and generative AI can and will supply the world with more bad information about what to eat, but those words will be meaningless if no one is there to consume it.
Footnotes
3 Recent posts from my Substack
3 Highlights from my current reading list
The World’s Broken Food System Costs $12.7 Trillion a Year by Matt Reynolds - WIRED
Walmart’s ‘Regenerative Foodscape’ by Lisa Held - Civil Eats
AI Deepfakes Are Making War in Ukraine Even More Chaotic and Confusing by Tony Ho Tran - Daily Beast
My email is mike@thefuturemarket.com for questions, comments, consulting, or speaking inquiries.