The Great Enabler

As I am used to saying, we are only humans, noisy, narrow-banded thinking machines that are prone to errors and failures of various types. If it had been possible for a human being to always think logically, remember a vast amount of data and reason consequently¹, we wouldn’t have needed a machine to supplement or support our thinking, but alas it is not the case. It turns out that we need a machine not only to store bits of data that our brain is incapable of storing, but also to aid deliberation over complex issues.

From the experience that I have with interacting with agents², vibecoding this website, having Claude as a copilot in my Obsidian, proofreading and correcting my text, helping in my research, etc, the machine taught me³ to think more clearly and actually disciplined my thinking process, forcing me to contemplate before actively prompting⁴. This made me come to a certain conclusion: that many⁵ problems that can be thought in advance and planned are notoriously difficult to think in advance and plan.

Not because humans “are only humans”, but because difficult problems possess one feature with which both humans and machines have issues. We don’t know what we don’t know. This series of essays might be a prime example of such unknown unknowns. When I started writing it, I didn’t know how I might finish it, what turns my thinking might take and where I land every time I started writing.

As with every scientific endeavour these notes are the result of trials and errors. Hell! The whole digital garden with various nooks and crannies is a by-product of my trials and errors of scientific thinking⁶. But recent development of my digital garden⁷ happened thanks to a machine that asked me “What is the question” type of question. Lifting the burden of the 85% of time that I previously spent getting into position to act⁸.

“What is the question” types of questions revealed flaws in the reasoning that I implemented in the PhD thesis of mine, provided unexpected solutions, and angles to attack the experimental data. Honestly speaking the attitude “what is the question” is well-known in scientific discourse, but previously humans could only interact with humans, expecting input from colleagues who are equally intelligent, equally interested in the broached topic, equally motivated to provide the answer and equally altruistic to give a hand.

AI moves scientific method up a notch on the spiral even though the output of discoveries is much less than input of effort⁹. The machine in 2026 actually shoulders the burden of knowledge that required 85% of thinking power of an ordinary person, hence enabling the researcher to get into position to think much faster. As a friend of mine said after we have sent for publication an article, “I would have never set myself on doing this without AI, which took the weight of Python scripting off my shoulders. And what I never did before I accomplished in just a week”.

I for one would have never created this website, written three books in three years, or concluded any research altogether. All mentioned above were domains of knowledge unknown to me, where in the pre-AI age I had neither experience nor expertise. Achieving anything¹⁰ would have required months of study before taking the first tentative actions¹¹. Studying and achieving something where a person doesn’t have expertise can be in itself rewarding but this kind of work is accompanied by errors, failures and other unpleasant stuff such as self-doubting, etc.; AI helped me, enabling the potential that I didn’t previously know I had.

Just recently I thought that I had worked out the experiment and actually know what results I should have gotten, but after putting the data through an analyzing pipeline I got disastrous for my PhD results. Back in pre-AI age it would have been an excuse for self-depreciation, long and painful communication with Uni Counsel, with whom I would have needed to align calendars first¹². But today I called in Claude instead of my scientific counsel and in two hours of brainstorming in the spirit of the “What is the question” type of questions I got myself a still tangible path out of failure.

I’ve been into science for slightly more than a decade and still think of myself as an inexperienced scientist, young and full of myself. A scientist to who Claude became a good advisor and friend who helped me understand the failures¹³ in the domain of knowledge that I had no or limited previous experience. AI is a great enabler of my work and enabler of slow scientific reasoning.

ChatGPT, Gemini, Claude, KIMI, QWEN and the rest of the lot drove the cost of coming up with new (not necessarily fresh) ideas, new hypotheses, almost to zero, enabling a fledgling researcher such as myself to play with data safely and affording ample room for trial and error.

Tycho and Kepler

Tycho Brahe and Johannes Kepler were the inspiration for this series of essays, I’ve read their story in “Tycho and Kepler: The Unlikely Partnership That Forever Changed Our Understanding of the Heavens” written by Ferguson Kitty, and at the moment of reading I had been imagining Kepler who snuck his way into Tycho’s circle of trust as a Solid Snake and Tycho as a Liquid Snake¹⁴ who wanted to keep the data accumulated over ten years within his family and in secrecy¹⁵.

Tycho Brahe¹⁶ (1546-1601) was a Danish nobleman, who studied the movement of celestial objects in the pre-telescopic era. He is known for the building of Uraniborg¹⁷ on the island of Hven given to him by King Frederick II. The cost of building was staggering, by some accounts around 1% of whole Denmark’s crown revenue^{18 19 20}. For decades Tycho used his naked eyes and numerous tools (mural quadrants, sextants) for observation of the movement of celestial bodies. Over decades he accumulated the data and managed to compile the most precise dataset in history²¹, especially on the movements of Mars.

The irony was, that Tycho didn’t possess the needed competences²² to read his own data. His belief engendered a hybrid understanding that encompassed religious dogma and scientific observation. From observation he had empirical gold, and that he put on a religious foundation²³. According to his cosmology²⁴ planets orbit the Sun and the Sun orbits the stationary Earth²⁵. What made things worse²⁶ is the fact that Tycho was a highly secretive guy, who grudgingly shared the data with others. In 1599 he became a royal mathematician to Rudolf II in Prague, and in 1600 hired as help a poor, but brilliant assistant.

Johannes Kepler (1571-1630), a genius mathematician who pursued the idea of geometric design of God’s cosmos²⁷. He was the direct opposite of Tycho, no money, no status, no data to prove his theories. In 1596 he came up with the nested-Platonic-solids model^{28 29} which was beautiful but wrong, something only data from Tycho’s decade of observation could have proved.

When they met, and Tycho acquainted himself with the mathematical prowess of Kepler, he set him to deal with an “impossible task”: calculating the orbit of Mars, the most unpredictable celestial object that broke the circular model³⁰. Kepler readily jumped at the task that he called “The war on Mars”.

In 1601 Tycho died and Kepler had hidden the data that Tycho accumulated his whole life, the data that was priceless for science of 17th Century. The heirs came to court but Kepler managed to wrangle it nonetheless, and finally he hit gold.

Even after Tycho saw the genius of Kepler he still provided Kepler with limited access to data and the death of Tycho maybe for the first time in years opened to Kepler the full trove and enabled him to fit Tycho’s Mars observations within 8 arcminutes³¹.

For anybody else 8 minutes would be a rounding error, but not to Kepler who trusted Tycho’s findings. Kepler assumed, judging by Tycho’s instruments, that 8 minutes is an accurate estimation, not the error, hence he made a claim that circular movement of celestial bodies is wrong.

Later that quote³⁴ gave Newton enough empirical floor, which later was used by Einstein.

So, here we are. In this part of the essay I would like to make a comparison of the Tycho-Kepler story to the modern age and technology.

To me AI is Tycho, the Great Enabler that shoulders the burden of grunt work. AI, like Tycho, spends time “at the quadrant and sextant”, accumulates data across the globe, doing, I think, more than 85% of manual labour. And the average researcher is Kepler, the funny thing is this Tycho only had the data, as AI and Internet have all the knowledge of the world, but he still came to the wrong conclusion³⁵. Every researcher with access to the full trove of Tycho’s data would still have to “wage war on Mars” and be the one to account for errors that everybody else did³⁶.

This story teaches us one important lesson. Ideas have never been the bottleneck. Kepler, like many of us, had them aplenty; he lacked data to test them. Tycho served his purpose and the cost of the data was ruinous: a king’s treasure and one man’s whole life. The AI drives this cost to zero. With the lifetime of billions of people’s data already in hand it enables millions of researchers to conduct trial and error research almost free of charge.

In pre-AI age the scarce thing was never the hypothesis but it was getting into position to test one.

The new constraint

Generating ideas costs nothing, data costs nothing, testing a hypothesis costs nothing, if all this is true, then where is the catch?

Judgement³⁷ …

Knowing what to trust and how to test is the only restriction that is left. If AI, as Tycho, hands the data freely to everyone in need, not every scientist possesses Kepler’s stubbornness to work with the data in a certain way.

BIO