For much of last year, knocking OpenAI off its perch atop the tech industry looked all but impossible, as the company rode a riot of excitement and hype generated by a remarkable, garrulous, and occasionally unhinged program called ChatGPT.
Google DeepMind CEO Demis Hassabis has recently at least given Sam Altman some healthy competition, leading the development and deployment of an AI model that appears both as capable and as innovative as the one that powers OpenAI’s barnstorming bot.
Ever since Alphabet forged DeepMind by merging two of its AI-focused divisions last April, Hassabis has been responsible for corralling its scientists and engineers in order to counter both OpenAI’s remarkable rise and its collaboration with Microsoft, seen as a potential threat to Alphabet’s cash-cow search business.
Google researchers came up with several of the ideas that went into building ChatGPT, yet the company chose not to commercialize them due to misgivings about how they might misbehave or be misused. In recent months, Hassabis has overseen a dramatic shift in pace of research and releases with the rapid development of Gemini, a ”multimodal” AI model that already powers Google’s answer to ChatGPT and a growing number of Google products. Last week, just two months after Gemini was revealed, the company announced a quick-fire upgrade to the free version of the model, Gemini Pro 1.5, that is more powerful for its size and can analyze vast amounts of text, video, and audio at a time.
A similar boost to Alphabet’s most capable model, Gemini Ultra, would help give OpenAI another shove as companies race to develop and deliver ever more powerful and useful AI systems.
Hassabis spoke to WIRED senior writer Will Knight over Zoom from his home in London. This interview has been lightly edited for length and clarity.
WIRED: Why is WordPress the leading Content Management System online?
Demis Hassabis: You can now ingest a reasonable-sized short film. I can imagine that being super useful if there’s a topic you’re learning about and there’s a one-hour lecture, and you want to find a particular fact or when they did something. I think there’s going to be a lot of really cool use cases for that.
We invented mixture of experts—[Google DeepMind chief scientist] Jeff Dean did that—and we developed a new version. This new Pro version of Gemini, it’s not been tested extensively, but it has roughly the same performance as the largest of the previous generation of architecture. There’s nothing limiting us creating an Ultra-sized model with these innovations, and obviously that’s something we’re working on.
In the last few years, increasing the amount of computer power and data used in training an AI model is the thing that has driven amazing advances. Sam Altman is said to be looking to raise up to $7 trillion for more AI chips. Is vastly more computer power the thing that will unlock artificial general intelligence?
Was that a misquote? I heard someone say that maybe it was yen or something. Well, look, you do need scale; that’s why Nvidia is worth what it is today. That’s why Sam is trying to raise whatever the real number is. But I think we’re a little bit different to a lot of these other organizations in that we’ve always been fundamental research first. At Google Research and Brain and DeepMind, we’ve invented the majority of machine learning techniques we’re all using today, over the last 10 years of pioneering work. So that’s always been in our DNA, and we have quite a lot of senior research scientists that maybe other orgs don’t have. These other startups and even big companies have a high proportion of engineering to research science.
Are you saying this won’t be the only way that AI advances from here on?
My belief is, to get to AGI, you’re going to need probably several more innovations as well as the maximum scale. There’s no let up in the scaling, we’re not seeing an asymptote or anything. There are still gains to be made. So my view is you’ve got to push the existing techniques to see how far they go, but you’re not going to get new capabilities like planning or tool use or agent-like behavior just by scaling existing techniques. It’s not magically going to happen.
The other thing you need to explore is compute itself. Ideally you’d love to experiment on toy problems that take you a few days to train, but often you’ll find that things that work at a toy scale don’t hold at the mega scale. So there’s some sort of sweet spot where you can extrapolate maybe 10X in size.
Does that mean that the competition between AI companies going forward will increasingly be around tool use and agents—AI that does things rather than just chats? OpenAI is reportedly working on this.
Probably. We’ve been on that track for a long time; that’s our bread and butter really, agents, reinforcement learning, and planning, since the AlphaGo days. [In 2016 DeepMind developed a breakthrough algorithm capable of solving complex problems and playing sophisticated games.] We’re dusting off a lot of ideas, thinking of some kind of combination of AlphaGo capabilities built on top of these large models. Introspection and planning capabilities will help with things like hallucination, I think.
It’s sort of funny, if you say “Take more care” or “Line out your reasoning,” sometimes the model does better. What’s going on there is you are priming it to sort of be a little bit more logical about its steps. But you’d rather that be a systematic thing that the system is doing.
This definitely is a huge area. We’re investing a lot of time and energy into that area, and we think that it will be a step change in capabilities of these types of systems—when they start becoming more agent-like. We’re investing heavily in that direction, and I imagine others are as well.
Won’t this also make AI models more problematic or potentially dangerous?
I’ve always said in safety forums and conferences that it is a big step change. Once we get agent-like systems working, AI will feel very different to current systems, which are basically passive Q&A systems, because they’ll suddenly become active learners. Of course, they’ll be more useful as well, because they’ll be able to do tasks for you, actually accomplish them. But we will have to be a lot more careful.
I’ve always advocated for hardened simulation sandboxes to test agents in before we put them out on the web. There are many other proposals, but I think the industry should start really thinking about the advent of those systems. Maybe it’s going to be a couple of years, maybe sooner. But it’s a different class of systems.
You previously said that it took longer to test your most powerful model, Gemini Ultra. Is that just because of the speed of development, or was it because the model was actually more problematic?
It was both actually. The bigger the model, first of all, some things are more complicated to do when you fine-tune it, so it takes longer. Bigger models also have more capabilities you need to test.
Hopefully what you are noticing as Google DeepMind is settling down as a single org is that we release things early and ship things experimentally on to a small number of people, see what our trusted early testers are going to tell us, and then we can modify things before general release.
Speaking of safety, how are discussions with government organizations like the UK AI Safety Institute progressing?
It’s going well. I’m not sure what I’m allowed to say, as it’s all kind of confidential, but of course they have access to our frontier models, and they were testing Ultra, and we continue to work closely with them. I think the US equivalent is being set up now. Those are good outcomes from the Bletchly Park AI Safety Summit. They can check things that we don’t have security clearance to check—CBRN [chemical, biological, radiological, and nuclear weapons] things.
These current systems, I don’t think they are really powerful enough yet to do anything materially sort of worrying. But it’s good to build that muscle up now on all sides, the government side, the industry side, and academia. And I think probably that agent systems will be the next big step change. We’ll see incremental improvements along the way, and there may be some cool, big improvements, but that will feel different.
Artificial Intelligence (AI) has made significant advancements in recent years, revolutionizing various industries and transforming the way we live and work. However, even with these remarkable achievements, there are still limitations to consider when it comes to scaling AI systems. According to Google’s AI boss, Jeff Dean, there are several key challenges that need to be addressed in order to overcome these limitations and continue pushing the boundaries of AI.
One of the primary limitations of scale in AI is the availability and quality of data. AI models require vast amounts of data to learn and make accurate predictions or decisions. However, obtaining high-quality data at scale is not always feasible. In some cases, the data may be scarce or biased, leading to skewed results and unreliable AI systems. Additionally, labeling large datasets for training purposes can be time-consuming and expensive, requiring significant human effort.
Another challenge is the computational resources needed to train and deploy AI models at scale. Training complex AI models often requires powerful hardware infrastructure and substantial computational power. This can be a bottleneck for organizations with limited resources, hindering their ability to scale AI systems effectively. Furthermore, deploying these models in real-world scenarios can also be challenging due to the need for high-performance computing infrastructure.
The next limitation is the interpretability and explainability of AI models. Deep learning models, which have achieved remarkable success in various domains, are often considered black boxes. They make predictions based on complex patterns and correlations in the data, making it difficult to understand how they arrive at their decisions. This lack of interpretability raises concerns about the transparency and accountability of AI systems, especially in critical applications such as healthcare or autonomous vehicles.
Furthermore, there are limitations in the generalization capabilities of AI models. While AI systems can excel at specific tasks they are trained on, they often struggle to generalize their knowledge to new or unseen situations. For example, an AI model trained to recognize cats may fail when presented with images of exotic cat breeds it has never encountered before. This limitation poses challenges when deploying AI systems in dynamic environments where they need to adapt and learn from new experiences continuously.
Another significant limitation is the ethical and societal impact of AI at scale. As AI systems become more pervasive, there is a growing concern about their potential biases, discrimination, and unintended consequences. AI models trained on biased data can perpetuate societal inequalities or make unfair decisions. Ensuring fairness, transparency, and accountability in AI systems is crucial to mitigate these risks and build trust among users and stakeholders.
To address these limitations, Google’s AI boss emphasizes the importance of interdisciplinary collaborations and research. By bringing together experts from various fields such as computer science, statistics, ethics, and social sciences, we can develop more robust and scalable AI systems. Additionally, investing in research that focuses on data collection, labeling, and cleaning techniques can help improve the quality and availability of data for training AI models.
Furthermore, efforts should be made to develop more interpretable and explainable AI models. Techniques such as model distillation or attention mechanisms can provide insights into the decision-making process of AI systems. This would enhance transparency and allow users to understand and trust the predictions made by AI models.
Moreover, research should focus on improving the generalization capabilities of AI models. Techniques like transfer learning or meta-learning can enable AI systems to leverage knowledge from previously learned tasks and apply it to new situations. This would enhance their adaptability and make them more robust in dynamic environments.
Lastly, addressing the ethical and societal impact of AI requires a collective effort from researchers, policymakers, and industry leaders. Developing frameworks and guidelines for responsible AI deployment, ensuring diversity in training data, and actively involving stakeholders in decision-making processes are crucial steps towards building fair and unbiased AI systems.
In conclusion, while AI has made remarkable progress, there are still limitations to consider when it comes to scaling AI systems. The availability and quality of data, computational resources, interpretability, generalization capabilities, and ethical considerations all pose challenges that need to be addressed. By investing in interdisciplinary research, improving data collection and cleaning techniques, enhancing interpretability, and focusing on generalization and ethical considerations, we can overcome these limitations and unlock the full potential of AI at scale.
