“I am not interested in hearing ridiculous Utopian rants.”
said one reader of yesterday’s Finxter newsletter on my note on declining cost curves and the implied disruptions.
I’ll include a screenshot of the email here:
💡 Info: Declining cost curves refer to the trend of production costs decreasing over time as technology and production processes improve. This trend leads to more affordable and accessible products and services, creating new opportunities and disrupting traditional industries and jobs. Think of it like a “cost savings snowball” that gets bigger and bigger as technology advances.
I got a lot of positive comments, too, but let’s focus for a moment on this one because it demonstrates how not to behave in the upcoming years of radical change and transformation.
Let’s recap my note on the declining cost curves for AI training:
ChatGPT and a ~60% annual reduction of AI training costs. The cost of average intelligence is now 10,000x lower, disrupting white-collar labor markets, logistics, universities, law, coding, artists, secretaries, and many other industries.
In this article, allow me to give you more data on this. Feel free to skip ahead if you’re convinced the previous statement is not “ridiculous”.
The following chart from ARK’s research shows the 10x decline in AI costs in only two years.
Declining by 70% twice is represents a total decline of 91%, i.e., slightly more than 10x!
This steep decline in AI training costs combines software and hardware training costs. Both sit on independently declining cost curves in their own right:
Grasping exponentials is tough for our biological brains.
With abundant intelligence, we’ll integrate super-human AI into more and more apps and devices. Everything will become more intelligent all around us.
Your coding IDE, Excel spreadsheet, heating system, car, glasses, clothes. Very soon, all of them will possess super-human intelligence due to the declining AI training costs.
As a researcher, I learned that I needed to look at the evidence rather than letting my biases cloud my view—facts, not opinions.
I am puzzled that billions of people today still don’t believe the evident, apparent, mind-boggling disruptions. Your job is as unsafe as mine. I could sugarcoat it, but what’s the use?
Look at the productivity improvement of AI-assisted coding:
Wow, just like that, an organization like Google, Tesla, Facebook, OpenAI can double their coding productivity. But what about graphical design?
If I write that the costs of intelligence decreased by 10,000x, I mean that literally:
Genuine valuable content that would cost $100 in 2020 can now be purchased for less than $0.01 with ChatGPT.
And ChatGPT today – at IQ 147 reportedly – is the dumbest it will ever be. From here on out, it will become more and more intelligent and be at our service at a lower and lower price.
Yes, it will reach super-human level performance in coding. Yes, it will create beautiful content, music, art, stories, sales pages, research practically for free.
What is left for us humans to do?
Well, your guess is as good as mine. At least, let’s try to remain open-minded and use the technology daily so you won’t get left behind.
For example, you could use ChatGPT to create advanced apps that were previously impossible for a one-person coding startup. Now, you can disrupt the big guys with a small ChatGPT API call.
If you’re interested in learning how to integrate ChatGPT in your own Python web apps, feel free to check out our new academy course:
As with all our courses, this course comes with a personalized PDF certificate with your name to showcase your skills to clients or your future employer.
I’ll close today’s article with this concluding thought: humans tend to adapt very quickly if forced to. Just two months ago, most people would have pushed back on the thought that AI is able to create super-human-level art, code, or content.
Now, only a few people do.
In only two years, AI training costs will have been reduced by another 10x, and we’ll see generative videos and whole books being written in seconds.
We will see peer-reviewed ChatGPT research papers that managed to pass double-blind journal review systems.
Throughout human history, intelligence has been a limiting factor for our economy. It no longer is. You can choose to ignore this or you can choose to embrace change. The world, however, will move on either way.
I’m sending you much love, my friend!
While working as a researcher in distributed systems, Dr. Christian Mayer found his love for teaching computer science students.
To help students reach higher levels of Python success, he founded the programming education website Finxter.com that has taught exponential skills to millions of coders worldwide. He’s the author of the best-selling programming books Python One-Liners (NoStarch 2020), The Art of Clean Code (NoStarch 2022), and The Book of Dash (NoStarch 2022). Chris also coauthored the Coffee Break Python series of self-published books. He’s a computer science enthusiast, freelancer, and owner of one of the top 10 largest Python blogs worldwide.
His passions are writing, reading, and coding. But his greatest passion is to serve aspiring coders through Finxter and help them to boost their skills. You can join his free email academy here.