Will artificial intelligence cure us?

A recent statement by Nobel laureate Demis Hassabis that artificial intelligence will be able to cure all diseases in ten years is circulating on the Internet. How much truth is there in that? And what are the achievements of artificial intelligence in medicine - so far

2076 views 3 comment(s)
Photo: Shutterstock
Photo: Shutterstock
Disclaimer: The translations are mostly done through AI translator and might not be 100% accurate.

"I think one day we might be able to cure all diseases with the help of artificial intelligence," said Demis Hassabis (48), Nobel laureate and CEO of DeepMind, on the American news channel CBS News' "60 Minutes" show on April 20, 2025.

"The end of all diseases?" the host asked.

"It's within reach. Maybe even in the next decade, I see no reason why not," Hassabis replied.

Together with his American colleague, John Jumper, this Briton (father Greek from Cyprus, mother from Singapore) developed the artificial intelligence model AlphaFold2, which can predict the structures of almost all of the 200 million proteins known so far. Proteins in the human body perform a multitude of biological functions. If there is a disruption in their production, structure or function, diseases can occur. For this, they received the Nobel Prize in Chemistry in 2024.

Hassabis' artificial intelligence represents a revolution

The 3D structure of a protein can often be used to estimate its function, "because we don't know that yet for most proteins in the human body," biochemist and computer scientist Katarina Zweig, who heads the Laboratory for Algorithm Responsibility at TU Kaiserslautern-Landau, explains to DW.

If the function of a protein is known and the structure is seen to be altered in certain diseases - this could possibly be the cause of the disease, says Zweig. "Then a drug could be developed that prevents the protein from taking on the wrong structure."

It used to take an entire doctoral dissertation to identify, calculate and model the structure of a single protein. "That took three to five years. The artificial intelligence that Hasabis has developed is truly revolutionary."

Although the causes of disease cannot be reduced to a single factor, "there are numerous examples in which proteins play a key role," says Florian Geisler, a leading researcher at the Fraunhofer Institute for Cognitive Systems (IKS).

The potential for the application of artificial intelligence in medicine is enormous. "Artificial intelligence will enable things in the coming years that we cannot even imagine today," he says.

The long and expensive path to market for drugs won't change anytime soon

However, even in ten years, we will probably still not be able to cure all diseases, believes Professor Zweig, because it is not easy to clearly determine which of the many proteins causes a particular disease.

"There are also mutations that have abnormal 3D structures. These may statistically appear to cause disease symptoms, but are actually harmless."

Even when it is certain that a particular protein structure leads to a disease, there is a lengthy process before a drug reaches the market. "It has to be tested in clinical trials, which requires a sufficient number of patients, the trials have to be approved - so, in my opinion, it certainly won't happen that quickly."

Professor Zweig, however, warns that treating the disease, even with the help of artificial intelligence, will still require large financial resources.

"That's why, in the future, drugs will only be developed where there are enough patients who have enough money to be able to pay for those drugs later."

Artificial intelligence is not yet mature for most diagnoses

In general, only a small number of medical diagnoses can be made based on clear rules, such as diabetes, says Professor Zweig: "There is a very clearly defined threshold value and a clear measurement method - and then diabetes is diagnosed."

Most other diagnoses, however, require a lot of knowledge and experience. "I don't know of any AI system that can do this reliably enough today to replace doctors."

Florian Geisler also believes that decision-making about therapies will remain in the domain of humans for a long time to come. "Primarily for ethical and legal reasons."

And probably also because artificial intelligence systems are still, in Geisler's words, a kind of "black box": "Something is input, something is output, but it is not fully understood how that decision was made."

Katarina Zweig puts it this way: "We cannot observe a machine while it learns and makes a diagnosis. We cannot know whether it is doing so according to the criteria that we as humans would apply."

Where artificial intelligence is already helping in medicine

But - with diagnostics based on CT scans, with the help of artificial intelligence, it is much easier to spot possible pathological changes, says Florian Geisler.

Artificial intelligence can also help identify unforeseen side effects when combining different drugs and thus contribute to the optimization of therapies.

In addition, AI could relieve the burden on the healthcare system, he says: "AI-based systems could, for example, automatically summarize patient conversations and prepare structured reports for health insurers. This saves valuable time in the healthcare system. This is where AI will play a key role."

Bonus video: