Bill Gates said that this year we are witnessing the birth of a new era. Everything that happened before, according to the Microsoft co-founder, will seem as distant as when using a computer involved typing at a C:> prompt. And he is not alone in this. For example, if you ask Google CEO Sundar Pichai, he will say that he has begun a technological change more profound than even personal computing, the Internet or mobile phones. That’s why, yes, we’re going to say it: 2023 was the year of artificial intelligence.
ChatGPT was the trigger. Launched in November 2022, the OpenAI platform showcased the power of generative artificial intelligence to the world. It caused a sensation: within five days, the chatbot had more than a million users around the world. Two months later, his first 100 million.
Two races then began. That of the big technology companies, who rushed like never before to tell the market that they could do wonders with this technology. ChatGPT was followed by Google’s Bard and Microsoft’s chatbot for Bing. Very quickly, the fascination with conversational models gave way to the fever of instant image generation. Tools like Midjourney showed us, for the first time, a Pope modeling Balenciaga. And they left many of us wondering: what is real?
At the same time—or perhaps not as much—governments were racing to try to figure out how to control something with such apparent transformative power. In 2023, the UN created the first world body to discuss the implications of artificial intelligence. And a summit in November managed to bring together the United States and China—and another twenty other countries—in the same declaration, to recognize that there is a potentially catastrophic risk for humanity.
There was everything: from fear of killer drones to romances with computers. But beyond the big headlines, several more low-key groups of scientists achieved magnificent AI-related milestones in 2023. It wasn’t all ChatGPT. Therefore, here we tell you five great feats that this technology left us this year.
- Artificial intelligence and climate prediction in the midst of the crisis
In 2023, Google’s DeepMind subsidiary presented an artificial intelligence model with the ability to predict where a hurricane will make landfall. They called it GraphCast. In its tests, it managed to predict weather conditions up to 10 days in advance, more accurately and much faster than current standard models.
GraphCast outperformed the European Center for Medium-Range Weather Forecasts (ECMWF) model in more than 90% of 1,380 test areas. In predictions for Earth’s troposphere—the lowest part of the atmosphere where most climate events occur—GraphCast outperformed the ECMWF model for more than 99% of climate variables.
The Google DeepMind tool managed to predict where Hurricane Lee would make landfall in Canada, a powerful event recorded in September, three days before existing methods. Being able to warn more in advance is key so that authorities and populations can prepare better. In other words, it buys essential time to save lives.
In 2023, artificial intelligence demonstrated unprecedented capabilities to process enormous volumes of data, extract insightful knowledge and improve predictive models, highlights the World Meteorological Organization (WMO) of the United Nations. Key support at a time when the climate crisis is worsening: 2023 will end as the hottest year in history and without clear solutions from the world’s governments.
- Google and genetic decoding
Again, Google DeepMind. In 2023, the company launched artificial intelligence capable of determining whether millions of genetic mutations are harmless or can cause diseases. Researchers announced in September that they had identified 89% of all key variations. This new development could accelerate the diagnosis of rare disorders and assist in the design of new drugs.
The new program is called AlphaMissense. It puts the magnifying glass on those variations that are known as nonsense mutations. They are variants in the DNA code in which a single letter is misspelled. A typical human being has 9,000 such mutations throughout his or her genome. Many times they can be harmless, but this alteration is also associated with diseases such as cystic fibrosis or cancer.
There are about 71 million possible missense variants, but only 0.1% had been classified by human experts. Only in that tiny number of cases was it known whether they were benign mutations or whether they were capable of generating some pathology. Google DeepMind, with its invention, has the potential to completely transform genetic analysis.
- Walk again thanks to artificial intelligence
“I feel like a little child, learning to walk again,” said Gert-Jan Oskam, a 40-year-old Dutchman who was in a serious motorcycle accident. He suffered a spinal cord injury that left him quadriplegic. For 12 years, he underwent different therapies that helped him regain some movement in his arms. However, none of the efforts had helped him walk again.
Oskam was part of a scientific experiment, developed by the Federal Polytechnic School of Lausanne (EPFL), in which he received electronic brain implants. The devices wirelessly transmit his thoughts to his legs and feet, through a second implant located in his spine.
It is like “a digital bridge”, which allows you to connect the brain and your spinal cord, bypassing the injured sections. They used algorithms based on artificial intelligence, which allow a person’s movement intentions to be decoded in real time from brain impulses. Oskam can now stand and walk with the assistance of a walker or crutches.
EPFL also introduced this year a new spinal implant, based on the same system, that helps people with Parkinson’s resolve their walking difficulties. And the Kurage company is already testing in Europe a system of pants also powered by artificial intelligence, which is helping people with partial paralysis walk.
- A way to read minds
A team of scientists in Australia managed to convert a person’s thoughts into written words. They achieved this thanks to a helmet with sensors and an artificial intelligence model called DeWave.
The research was carried out by the GrapheneX Center at the University of Technology Sydney (UTS). The scientists asked 29 volunteers to silently read excerpts from a text while wearing a type of hat. The device allowed their brain activity to be recorded using an electroencephalogram (EEG). These signals were then decoded thanks to artificial intelligence.
Chin-Teng Lin, one of the project leaders and director of the GrapheneX Center, highlighted that they have already managed to increase the helmet’s accuracy level from 40% to more than 60%. But they are not the only ones working on solutions that allow “mind reading.”
A team at the University of Texas has developed a decoder that can translate thoughts into a continuous stream of text. According to researchers, they were the first to achieve something like this without the need for surgical intervention. Unlike the University of Technology Sydney helmet, this decoder works from fMRI (functional magnetic resonance imaging) scans.
And a group of Japanese scientists this year presented new technology that can read brain waves, identify what you’re thinking, and translate it into a very high-resolution image. It is so exact that it can even capture shapes and colors. Researchers at the Graduate School of Frontier Biosciences at Osaka University used Stable Diffusion, one of the most famous image generative artificial intelligence systems.
- Artificial intelligence and the extraterrestrial search
A group of scientists this year presented the “holy grail of astrobiology.” It is a system, powered by artificial intelligence, that recorded an accuracy close to 90% in its tests to evaluate past and present life samples.
The method has “the potential to revolutionize the search for extraterrestrial life and deepen our understanding of both the origin and chemistry of the earliest life on Earth,” explained Robert Hazen, of the Carnegie Institute for Science in Washington and co-director of the study.
The new test not only identifies a molecule or specific groups of compounds in a sample. The researchers pointed out that, thanks to artificial intelligence, they can differentiate between samples of biotic nature – plants, animals and bacteria – and of abiotic origin – water, soil and atmosphere. This analysis is achieved through the identification of subtle differences within the molecular patterns.
Another NASA group is training a new artificial intelligence that could be 88% more likely than us to find life on Mars. The objective of this technology is to help space rovers – such as Curiosity and Perseverance – to analyze the territory of the red planet faster and better.
These researchers combined ecological statistics with machine learning techniques. In this way, they gave artificial intelligence the ability to map “biosignatures”: any characteristic that provides evidence of past or present life.