Google has for years been the great dominator in the development of artificial intelligence (AI) systems. The acquisition in 2014 of deep mindthere to start up A London company has focused on AI research and developed a program, AlphaGo, capable of defeating a grand champion of Go, the complex Asian board game, and opening the debate on whether the capabilities of this technology would eventually surpass the human mind.
But Google’s quiet dominance was interrupted last year by another to start up, in this case OpenAI. The launch of ChatGPT, the most successful app in history, took major tech companies by surprise as they were forced to accelerate their AI programs. In April of this year, DeepMind, which until then operated as a relatively independent research laboratory, and Google Brain, the technology company’s other major research division, merged into a single organization, Google DeepMind, in which many of the world’s best scientists in this discipline.
Colin Murdoch (Glasgow, 45 years old) is the commercial director of this new super AI division of Google which has just presented its first toy: Gemini, a multimodal generative AI platform capable of processing and generating text, code, images, audio and video. from different data sources. Those who have used it say it far outperforms the latest version of ChatGPT and once again puts Google in the fight to dominate this market.
An electronics and computer engineer by training, Murdoch joined DeepMind nine years ago, after gaining experience managing startups and large companies. Their job is to ensure that the AI advances made by Google’s science team ultimately impact the general public. He attends EL PAÍS from London via video call.
Ask. Is Gemini the definitive answer to ChatGPT? What new does it bring to the popular OpenAI application?
Answer. Gemini represents a significant step forward in the development of AI. It’s our largest and most powerful model yet: it understands and reasons through text, images, audio, video, and code to help people be more creative or learn. For example, let’s say your child brings home physics homework and needs help understanding what they did right and wrong. If you took a picture of the page, Gemini would not only give you the correct answer to the problem, but would read the letter and explain what the child did right, what he or she did wrong, and the concepts behind it. Users can also interact with Gemini through Bard, which now works with Gemini Pro and is more effective for understanding, summarizing, reasoning, coding and planning. It is already available in English in more than 170 countries, and in the coming months it will be accessible to billions of people through other core Google products such as Search, Ads, Chrome and Duet AI. In the long term, tools like Gemini will transform the way billions of people live and work around the world.
Q. What do you think of the commotion caused by the departure and then reinstatement of Sam Altman as CEO of OpenAI?
A. It’s been a very interesting few days in the industry. But we remain focused on our work of launching world-class products and research. We’ve had an incredibly busy few months: since the announcement of Lyria, our advanced AI music generation modelwhich will increase creativity and generate new forms of innovation for artists, creators and fans of the future, until the publication of GraphCast, our weather forecast system latest generation and Gemini. We are very confident in our technology portfolio and are excited about the year ahead.
Q. What is artificial intelligence capable of today?
A. Our research lab seeks to improve people’s lives, and I think AI is a good tool to do that. As long as you work with care, which is in DeepMind’s DNA. One of the areas I’m most excited about is what we call science at digital speed: AI helping to advance science. I am going to give you an example. Proteins are the building blocks of cells. When they malfunction, they can cause problems or illnesses. Science has therefore been studying the structure of proteins for years, in particular the shape of these structures, which really tells us how they work. The number of forms that these structures can acquire exceeds the number of atoms in the universe. Well, two years ago, a DeepMind team managed to develop an algorithmic model, AlphaFold, capable of determining the structure and appearance of amino acids, the essential building block of proteins. We know the structure of 200 million proteins and we have discovered this knowledge. Our tool is used to accelerate research into methods to combat antibiotic resistance. It is also used to accelerate research into enzymes that consume plastic in the oceans. And in cancer vaccine research.
Q. Have they progressed in other areas?
A. We hope that progress will be made in nuclear fusion, the cleanest energy source available. To achieve this, plasma flows through superconductors at high speeds, around 10,000 revolutions per second. Magnets are used to reduce plasma friction in the tubes. We use AI to try to optimize, in real time, the calibration of these magnets so that the resistance is as low as possible.
Q. The interest of the general public, and apparently also of businesses, has turned to generative AI. Do you think this could harm progress in many other areas of AI, like the ones you just mentioned?
A. We have been working on generative AI for a long time. In fact, the models that are successful today are based on an architecture called Transformer that Google scientists developed five years ago. What’s happened over the last 12 or 18 months is that things have escalated very quickly: we have bigger models and more data. The fundamental change is that we can identify with these models of conventional language, like the one you and I are talking about now, and that makes it more accessible. Before, only computer scientists could relate to this technology; now anyone who can speak and write.
AI is very powerful and promising, but you have to be very careful because it is a very powerful technology
Q. You mentioned the importance of being careful when developing AI. What kind of rules do they follow?
A. AI is very powerful and promising, but you have to be very careful because it is a very powerful technology. We have a number of operational principles on how we can conduct our investigations. A second element is that we conduct research ourselves in areas such as bias and equity, to ensure that we are addressing these challenges properly. Third, it is important to have a correct institutional setup in the organization and an appropriate culture. We have multidisciplinary groups including ethicists, engineers and a wide range of professionals from different specializations, who test and analyze the benefits and risks of each system we develop. We also invite external specialists to help us.
Q. How do you think this technology should be regulated?
A. Regulation is important. I think this needs to be measured and proportionate so as not to restrict innovation and at the same time mitigate significant risks, because I think this is an exceptionally promising technology.
Q. Do you think the approximation of regulations on artificial intelligence that the EU has just approved is correct?
A. I think so, it establishes a proportionate and risk-based approach to each tool. It seems to me that this is a good starting point for the global debate. It’s important that we try to promote this kind of coordinated approach to regulation and policy around the world so that we can maximize the benefits for all – and there are many – and we can also adequately mitigate their risks.
Q. DeepMind was until now Google’s advanced AI laboratory. Has this changed after your company integrated with Google? Should they now direct their work more towards business outcomes?
A. I think the merger was a very successful initiative. On the one hand, we have a scientific team without equivalent in the field of AI; on the other, a huge market that we can access thanks to Google, which gives us the opportunity to try to solve people’s problems. My job is to find ideas at the intersection of these two spheres. And when we find them, let’s incubate each idea and move it forward.
Q. Give me an example in which these two elements coincided.
A. At Deepmind we have software, MuZero, capable of playing chess, Go and other complex games. One day while talking to someone from YouTube, they told us that they need to reduce the bandwidth needed to deliver our videos to people around the world, so you can watch them no matter how fast your connection is. Internet. There was a very creative moment where we realized that a video, in essence, is like a game of chess: it can be seen as a succession of individual still images, and there are transitions between these images . Each of these images can be a position on the board, and the transitions, chess moves. So we applied MuZero to a video and gave it the goal of reducing its size, compressing it. We found that this had a huge impact on the weight of these videos, and now this technology is already integrated into YouTube.
Limited time special offer
Subscribe to continue reading
Read without limits