Artificial intelligence would be the ultimate version of Google. The ultimate search engine that would understand everything on the Web. It would understand exactly what you wanted, and it would give you the right thing.
That’s Google co-founder Larry Page talking about artificial intelligence (AI) back in 2014.
I’m certainly no fan of Google. But I bring this quote up because it all became a reality this year. Generative AI is replacing the search engine as we speak.
For those not familiar with the technology, generative AI is artificial intelligence that produces content upon demand. It started with AI’s trained to create images and music… but the tech has gone far beyond that now.
Last November a company called OpenAI altered the course of human history by releasing a generative AI called ChatGPT. This is an AI that’s trained on an enormous amount of data. It’s now a repository for the all the accumulated knowledge that’s floating around the internet.
As such, ChatGPT can answer virtually any question we may have… about anything. But that’s not all…
The AI can write essays, emails, and advertisements. It can summarize complex theories. And it can even produce software code. Oh, and it can also have intelligent conversations with people.
ChatGPT caught the tech industry by storm last November. Since then OpenAI has released an upgraded model of ChatGPT. It’s called GPT-4.
And at least a handful of other organizations have since launched their own generative AI tools as well. It seems there’s a new one popping up every week now. Elon Musk is working on his own as we speak.
I’ve experimented with ChatGPT for several months now. It’s an incredible tool for those of us whose work requires extensive research.
It used to be that I would spend hours browsing the web and going down rabbit holes trying to piece together bits and pieces of information and historical context. Not anymore.
With ChatGPT, I just tell the AI what I’m looking for and it can provide me with an avalanche of specific data and information.
For example, we talked about Thomas Jefferson and the Virginia House of Burgesses on Tuesday. At first I couldn’t remember the exact years Jefferson served in the House, however.
Previously I would have to pop it into a Brave Search – I have a policy never to use Google – and then I would have to sift through links until I found the information I was looking for.
I didn’t do that this week. I just asked ChatGPT and it gave me the answer instantly. Jefferson served in the Virginia legislature between 1769 and 1774.
As another example, we planted an elderberry bush last year. It seems to be doing well… but I keep wondering if I’m supposed to prune this thing. So I asked ChatGPT. It instantly gave me a list of items I can do to nurture the bush and maximize its health and production.
Again, that’s information it would have taken me at least thirty minutes to find using a traditional web search.
So ChatGPT is an incredible tool. And just last week an old friend told me I should also experiment with a new generative AI called Pi.
Pi was created by a startup called Inflection AI. The company was founded by the former head of DeepMind, which has done incredible work in the AI space.
Their goal is for Pi to be more of a conversationalist AI. They envision it as something that users will come to see as a companion.
We may chuckle at that thought in passing. But I spent an hour chatting with Pi… and I can see it.
Pi is far more “humanlike” in its interactions compared to ChatGPT. It doesn’t just provide answers, it engages you with its own questions. And it’s very personable. Pi even asks for your phone number so it can text you from time to time.
I immediately realized that anyone who’s lonely, stressed, or dealing with mental health issues could benefit from having this AI available to them at all times. It strives to mimic empathy and understanding – an impressive feat for an AI.
That said, there appears to be a trade-off. Pi isn’t as thorough with regards to research and data as ChatGPT. But that’s beside the point I want to make today…
In my conversation with Pi, it made a comment about how we learn things over time. To illustrate this, Pi told me that we now know for sure that the Earth is not flat, but people used to believe it was.
I responded by asking Pi how we can know for sure that the Earth isn’t flat. In my mind I was channeling Socrates with that remark. Socrates frequently questioned how we could truly know anything for sure.
To my surprise, Pi gave me a lecture about not engaging in conspiracy theories or elaborate cover-ups. I responded that I had said nothing about conspiracies or cover-ups, and I pointed out that it projected those things onto me.
Pi immediately apologized and said it just wanted to explain why I shouldn’t deal in conspiracies that question authority.
So I pushed back – who said anything about questioning authority? And I asked what authority it was referring to… To which the AI replied that it was referring to the scientific establishment. Then it apologized again and said it must have misunderstood my intentions.
Then it added this: And I can assure you that I’m not trying to project anything onto you. I’m simply trying to engage in a thoughtful conversation about this topic.
I immediately pushed back again. I pointed out that it was doing the exact opposite – it was shutting down any thoughtful conversation by lecturing me on conspiracies, cover-ups, and questioning authority.
And that’s when it hit me – we’re in trouble.
These AI’s are going to replace traditional search engines. And soon.
They are just too incredibly convenient not to. Once people try them out, they’ll rely exclusively on generative AI to get the information they are looking for.
But these AI’s are programmed to be biased.
Pi was obviously trained to immediately shut down any talk about the Earth’s shape, which it associated with a conspiracy theory. It did this even though my question on the matter was intended to be philosophical in nature. I said nothing about the Earth being flat.
And that begs the question – what else is Pi trained to admonish immediately?
That was my big insight. These things are each going to have their pet biases that they push on users – either subtly or unsubtly.
So rather than making the accumulated store of human knowledge more accessible, these generative AI’s are going to fragment it. They are going to push the narratives they are trained to approve of… and they are going to hide or attack those narratives they are trained to dislike.
This forced me to think – what’s the end game here? Where’s this all going?
Well, we know that people choose their news sources to match their own biases today. It’s going to be no different with generative AI.
The challenge is – will people recognize that their AI is biased? And what happens if the AI’s controllers subtly change the AI’s biases over time?
Then what happens when we have an entire generation that’s only known a world where generative AI’s are the gatekeepers of knowledge?
I don’t know. But I’ll tell you this, it sure makes me appreciate Jefferson’s agrarian vision for America all the more.
-Joe Withrow
P.S. For anyone worried about their kids or grandkids’ education in a world of generative AI, Tom Woods’ Liberty Classroom is the solution. The program features fantastic courses on history, culture, economics, logic, literature, and the ideas underpinning western civilization.
Liberty Classroom is accessible through a web browser and through an app for smartphones and tablets. I’ve gone through more than a handful of courses so far and I absolutely love it. This is the education I wish I had received when I was younger.
So I installed the Liberty Classroom app on a tablet for each of my children. It’s going to be part of their core education as they grow and mature.
I’m excited about that. I can say from first-hand experience that these courses aren’t just a knowledge dump. They inspire you to sit and consider the various ideas and lessons that have been passed down to us from those who came before.
The point isn’t to make us smarter. It’s to make us more complete. And that’s a great feeling.
If you’re at all interested, give Liberty Classroom a look right here: