Created by OpenArt
In my quest for a better understanding of artificial intelligence so far I have been listening either to the academics or the developers of the technology. In other words, to the people who are deeply involved and are inside the processes. It was different this time – Yuval Noah Harari is a historian and an anthropologist who has the privilege of looking at AI development from the outside.
Here are my three takeaways after watching his conversation with Kara Swisher, an American journalist, about information, truth and artificial intelligence on YouTube:
Lesson #1
All information is not the truth. Truth is just a small subset of information. Truth is costly whereas fiction is cheap because it doesn’t require research or verification. Truth may be too complicated to digest easily, whereas fiction is usually simple. Truth can be painful sometimes, whereas fiction can be easily made as attractive as needed.
It was not the Gutenberg’s printing press that triggered the scientific revolution. It took two hundred years from the 15th to 17th century to get there. In between the European continent experienced some of the worst wars to date and unprecedented witch hunts.
With the help of printing presses the old stories about the odd elderly ladies making potions and healing bodies and souls suddenly were lumped together into a comprehensive theory of “global conspiracy led by Satan with agents present at every village”. It gained momentum quickly and was strengthened with the claim that it was all evidence-based – “I heard from some people that…”.
It required building new institutions to sift through all the information to help to stop this. It was one of, if not the first, victories of truth over commercial success.
Lesson #2
Artificial intelligence means that it learns and changes by itself. If a machine can’t learn and change by itself it is not artificial intelligence.
The new word “agentic” is becoming increasingly popular among the tech people. It is used to describe someone who behaves like an agent or has the properties of an agent. An AI agent is an entity that can:
- make decisions by itself that you can’t anticipate;
- invent new ideas that you can’t anticipate;
- learn and change.
In the first generation AI agents create what is mostly similar to what humans have produced so far (they have absorbed everything that has been created before). Over time we can expect artificial intelligence to get more creative but also more alien. We are going to face more and more of an alien culture from a non-organic entity.
The financial system is an excellent playground for artificial intelligence – it is only about data. It is much more difficult to build self-driving cars because you have to consider different human behaviours, stray dogs and animals, potholes on the road etc. Data, in comparison, is easy.
Lesson #3
Humans are historically very good at adapting to change if they are given sufficient time. It does not seem though that we can slow the AI development down. It leaves just the option to introduce urgent rules:
- corporations should be made liable for their algorithms, not for the content their users post – it is the algorithms that promote certain posts; the algorithms should not be driven just by a single objective to increase user engagement;
- the use of artificial intelligence in impersonating humans should be banned;
- an international institution should be established to monitor the process and to tell the people what is happening.
We have thousands of years of experience in dealing with humans and no experience in dealing with artificial intelligence. This suggests that human moderation should be more trusted.
And, as a final note, historically the editors of major newspapers have played an extremely important role in driving the processes in the society by shaping the information agenda. There are several known cases in history which demonstrate that there has been an established career path from journalist to editor to dictator. Today editor is among the first functions to become automated. And the instructions to the algorithms are simple – increase user engagement.
For those willing to try and pick top lessons of their own here is a link to the video and here – to Yuval’s latest book “Nexus: A Brief History of Information Networks from the Stone Age to AI”. And yes, I may receive a commission if you decide to buy it using this link.
* from anything that you are reading, watching or hearing you can realistically expect to remember only a limited number of things. My solution is to pick just 3 items or ideas from any material. This number is non-negotiable. Even the most extraordinary experience gets compressed into 3 things to remember. This approach has worked well for me.
This note was first published on medium.com on 3 November 2024.
Aivars Jurcans has more than 20 years of corporate finance and investment banking experience. His services are currently available through Murinus Advisers. More of his writings can be found on his page Corporate Financier’s Notes.
