- Watch your fingers when we start it, it hasn't read Asimov's robot laws, so it doesn't know it can't hurt you, says Professor Ross D. King with a lively wink behind his glasses.
The small room in the chemistry building at Chalmers is dominated by a glass cage where two robotic arms rule. All around them is full of lab equipment.
- That box in the middle is a fully automatic microscope, and those blue and red things can read light absorption and stuff. There you have a mass spectrometer, King shows.
All of these are parts of the research robot Eve, which can write hypotheses, design and conduct experiments, and draw its own conclusions from the results.
- I don't think it's possible to fully understand biology and medicine if we can't do many times more experiments than today. Although I can't plan thousands of experiments in one day, a research robot can, says Ross D. King.
The idea was born during the doctoral years in the late 1980s. King's thesis revolved around predicting the shape of proteins using artificial intelligence. But when he tried to find chemists who could assist with the experimental part of the research, no one took it.
- Then it hit me, what if a robot could do the experiments? says King.
In retrospect, he was on the right track. In 2024, the Nobel Prize in Chemistry went to the researchers behind Alphafold, an AI developed by Google Deepmind that cracked the code to predict the shapes of proteins.
- I might have been able to get a Nobel Prize, says Ross D. King with a small laugh. The difference was that they had at least a million times more data and a million times faster computers to work with.
Instead, he has dedicated the last 20 years to that other idea: can robots do this?
He named the first research robot Adam. It was able to perform independent analyzes of yeast cells, 1,000 experiments a day, and became the first robot to make its own scientific discoveries, attracting much attention after a publication in the journal Science in 2009.
Then came Eve. Inside the glass cage, its robotic arms move quickly.
Eve conducts experiments on molecules in search of drug candidates for what are commonly known as neglected tropical diseases; malaria, snail fever, dengue fever and others that the pharmaceutical industry tends to ignore because they are considered unprofitable.
- There you have an interesting example of how capitalism has failed, sighs Ross D. King.
Nobody is interested in curing a disease that mainly affects the poorest people on earth, he believes.
The most interesting discovery Eve has made so far is an antibacterial molecule called triclosan. It's not a particularly exotic substance, it was long an active ingredient in Colgate toothpaste, but in experiments Eve has shown that it can potentially treat malaria.
- But we did not get any grants to follow up on that research, we have tried to see if any pharmaceutical companies want to cooperate, but it is not a priority disease for them, says Ross D. King with noticeable irritation.
King's vision is that robots and AI can make this kind of research radically cheaper and faster. Something he believes is required to solve problems such as famine, disease, poverty.
- We need more advanced science and better technology, that's the only solution for even the poorest people to be well off, says Ross D. King.
He mentions cancer care. The most common treatments against cancer hit very wide and hard, but those who have enough money can nowadays pay dearly to have their individual cancer tumor analyzed and choose the most effective and mild treatment.
"Research robots could make that kind of technology cheap enough to be available to everyone, not just billionaires," says Ross D. King.
A few floors up in the house, Eve's successor Gensis is housed in a compact black box. This robot is designed to carry out careful experiments on yeast, and has recently given rise to a first publication in the journal Journal of the royal society interface.
- We use yeast to make bread and beer and all that, but in biology we use yeast as a model for human cells. Yeast cells and human cells look basically the same, says Ross D. King.
His research colleague Ievgeniia Tiukova breaks in.
- In biotechnology, yeast is also useful for producing lots of different chemicals, biofuels and medicines. Or as a resilient food source.
Gensis can grow yeast cells in an extremely controlled environment, and investigate how these are affected by external factors. When the robot is fully developed, around 20 black boxes will fill the lab and be able to run numerous simultaneous experiments, without much intervention from the researchers.
- With this robot, we can solve a bottleneck in biotechnology, says Ievgeniia Tiukova.
Although Ross D. King lost the chance to receive the Nobel Prize for his research on the shapes of proteins, he still holds the award as a yardstick. But now his robots are targeting it.
He is one of the instigators of The Nobel Turing Challenge, which aims to create a robot capable of producing Nobel Prize-worthy research (although robots cannot by law be awarded a Nobel Prize).
When the competition was established in 2020, the goal was to reach there by 2050. Ross D. King himself was doubtful if it was really possible, but just a couple of years later Chat GPT was released and the whole world started pouring money into AI development. Now he calls the goal "pessimistic", even though his own research robots are still dealing with relatively simple science.
- People don't immediately get better at research, but the machines do. So it's just an extrapolation, says Ross D. King.
He compares it to how a chess grandmaster plays the same game, according to the same rules, as a beginner. It's just about a deeper understanding, a different way of approaching the problem.
- In the same way, I don't think there are any fundamental differences between simple science and the top layer of science. So I see no reason why the research robots cannot reach there, he says.
Think your lab will make it first?
- It is difficult to compete with business, they have access to many times more money than we can ever get. But I also don't want Open AI or Google to take the lead in science.
