This may be a chore to follow, but you should, because I am going to take you on a deep dive down the rabbit hole of artificial intelligence and what it can do to our lives.
Or maybe what it has already done.
AI has made another incursion into my own life. I say “another” incursion because I already subscribe to Google Gemini and Chat GPT and Claude. But I opened Facebook this morning and was greeted by “My name is Meta AI. Think of me like an assistant who's here to help you learn, plan, and connect. What can I help you with today?”
My mind went blank. Well, not really, because I already knew from reading the tech sites to which I am drawn, and listening to the tech podcasts that help me go to sleep at night, that Meta was launching its version of artificial intelligence in all its products. And that means Facebook, Messenger, WhatsApp, Instagram, and Threads. So now it’s here.
Most people never stop to think that one company owns all these apps. And even if they do, they don’t ponder on whether this will affect them. But it might.
I wonder how those of us who use these apps are going to be trained to see reality.
Underneath the surface, all Meta’s apps are using artificial intelligence based on Meta’s LLM called Llama 3, which Meta has decided to make freely available. Llama3 has been trained on data we provided to Meta when we used its products, so now we at least get to see our own data provided back to us.
I don’t know about you, but Meta’s products have been woven into my life for about two decades. I use WhatsApp and Threads the most, but I do use them all. And so do most Americans.
These are big platforms, and they help fragment America.
But there are other platforms that may fragment us even more.
YouTube is owned by Google, and Google also has a LLM called Gemini, which has been trained on Google’s data. That’s the data we gave them when we used search, or Gmail, or any other Google app. YouTube by and large is the most widely used online platform measured in our survey. Roughly eight-in-ten U.S. adults (83%) report ever using the video-based platform.
While a somewhat lower share reports using it, Facebook is also a dominant player in the online landscape. Most Americans (68%) report using the social media platform.
Additionally, roughly half of U.S. adults (47%) say they use Instagram.
If you use Windows, you are using ChatGPT, which is based on the LLM from Open.ai
We can go deeper when we cut the data into smaller pieces.
For example, Hispanic adults (49%) are more likely to use Tik Tok than black adults(39%), and Twitter (X) is more likely to be used by black adults than white adults. Twitter has an AI called Grok, trained on the data from those who use it.
62% of 18- to 29-year-olds say they use TikTok, which is much higher than the share among adults ages 65 years and older (10%). We think TikTok has taken over, but it is only used by about a third of adults. Those, however, are the young ones.
I know it’s not fascinating to read about Pew Research surveys, but when you then think about us coming up on an election that is, in one way or another, a referendum on democracy, this kind of research deserves closer scrutiny.
Smarter data scientists than I should really take a look at who ( which segment of the population) uses what (social media platforms or AI models) in America. This kind of in-depth information could tell you a lot about why we are so divided and can’t talk to each other anymore.
It’s simple: we no longer have a common platform for exchanging information. This is bigger than Republicans or Democrats; it is differences in epistemology--how we know things.
If we are increasingly going to get our perceptions of reality from large language models training an AI that powers the social media apps we use, we may increasingly have different views of what is “true,” what is “real,” and what is “knowledge.”
Does your reality come from Quora (the LLM behind Claude), or Meta, or Google? Or perhaps China? (TikTok).
This could destroy democracy.
Ga-a-a-a-h. Please correct me if I am wrong.