THE RISE OF THE MACHINES: OpenAI & ChatGPT
By: J SCOTT
I normally use this spot to discuss things like real estate, investing, the economy, money or some related topic. But those of you who know me know that my first career was as a senior tech/product guy in Silicon Valley. In fact, I still sit on several advisory boards for technology startups and remain well entrenched in the industry.
Long story short, I'm a big geek at heart.
When I see new and interesting technology, I like to talk about it. (I still have people thanking me for the fact that I was writing about cryptocurrency and Bitcoin back in 2016. :-)
Over the last couple weeks, I've been playing with one of the most interesting -- and likely to be one of the most influential -- technologies I've seen in a long time. I'm not yet sure how it will impact real estate or investing, but I've been having a lot of great discussions about with some of the smartest engineers and product people I know about how this technology is likely to impact the world around us. Future iterations and improvements will likely fundamentally change big things -- from the global economy to perhaps even our perception of reality.
That's a pretty big buildup, I know. But for those of you who have already been playing with this tech over the last week or two, you're probably in agreement with me.
OpenAI / ChatGPT
The company is called OpenAI and the specific product is ChatGPT. It's the next step in the evolution of artificial intelligence, and the best way to describe the product is like a Google search engine but to be used more as a human chatbot that can process much more nuanced thought and interpretation of data and can communicate through natural language.
But, the more you understand it, the more you realize that the comparison to a Google search engine is actually not a very good one. Google is great at indexing data and matching search queries to content that has been generated by others. But, ChatGPT actually creates unique content. Content that is well structured and seems like it could have been conceived of and written by a human being.
For example, you could ask the bot to write you a 1,000 word essay on thematic ideas from Mary Shelly's book, Frankenstein (my high school senior paper topic). Using its extensive training set, it would create custom narrative that addresses this topic. And it would almost certainly do so better than my high school self did many years ago.
Ask it the same question again -- or ask it to expand on specific ideas it had already communicated -- and it would generate new content, presenting it in full English sentences, written better than most adults I know.
To give you an idea of the scope a couple examples of interesting queries I've done along with the results:
Now, if you play with it, you're likely to find that it's answers aren't always accurate. Especially for quantitative analysis, and sometimes even for very simple math.
Here are a few examples (though if you ask the question multiple times, the correct answer often is found, and I've seen the answers getting better over the past couple weeks):
And discussions aren't limited to a single question/answer. It's quite possible to have a seemingly normal conversation with it, where it can follow logic from one prompt to the next:
What Does It Mean For the World
Again, we're probably still several iterations of the software away from it being "smart" enough to have any fundamental impact on our day-to-day lives, but it's not difficult to imagine some ways that this could change our world -- potentially even in the short-term.
On the good side, this will give a voice to a lot more people. As someone who likes to teach and who is reasonably good at explaining things, I'm fortunate that I'm a decent writer. Not everyone is. And there are a lot of people out there who have a tremendous amount of knowledge to convey, but don't have the skills to convey them.
Technology like this will make it much easier for people who aren't great writers -- including those who want to write in a language that is not their first -- to convey their ideas. While ChatGPT likely isn't going to do a better job of explaining quantum mechanics than a physics professor, it could certainly assist that professor in writing a book on the topic by helping to organize ideas, improve grammar and syntax, and convert complex ideas into simpler ideas that the average reader could understand.
In many ways, I expect technology like this to have a similar impact on the world as the calculator and spreadsheets. As someone who grew up long enough ago that I remember when these things weren't omnipresent (we didn't have them in our pockets everywhere we went), I appreciate being able to analyze a rental deal or calculate an IRR without a few hours of work.
Technology like ChatGPT will do the same thing for the world of content. Putting answers to seemingly complicated -- or even impossible -- questions in the palm of our hands.
Of course, this is also going to have an impact on education. Just like calculators provided us a crutch that alleviated our need to actually learn basic math, I suspect tools like ChatGPT are going to have a negative impact on children learning to write and communicate effectively.
This is what one educator has to say on this: The End of High School English.
If kids (and adults) can avoid having to develop their writing and reasoning skills, will this contribute to the collective dumbing-down of civilization? While at the same time inflating our egos, having instruct a machine to communicate for us, but taking the credit ourselves. Or worse yet, much like Google has done, will this type of technology ultimately replace critical thinking and true research?
Most concerning, with the ability to generate nearly unlimited content in short periods of time -- content that will soon be indistinguishable from that created by a human -- we will likely be overrun by content. Imagine bots on social media that can write or comment the same idea thousands or millions of times, with each being uniquely worded and seemingly written by a different person.
Do millions of people agree on the topic? Or is one person making it look as if millions of people agree?
While there are lots of potential great applications for AI models like this, there's also plenty of opportunity for it to turn us lazy. And for the worst of us, allow us to exploit it for our own gains.
How Can You try It?
Want to try ChatGPT? It's easy!
Just go to OpenAI.com, create an account, and get started...