Data Science

A machine learning algorithm called CEBRA is unlocking the hidden structure in neural code, with the potential to revolutionize brain-machine interfaces (BMIs). The algorithm can be used to decode what a mouse sees while watching a movie, predict primate arm movements, and reconstruct the positions of rats as they roam. The research, led by Mackenzie Mathis, EPFL’s Bertarelli Chair of Integrative Neuroscience, was published in Nature. Continue Reading →

The White House announced AI-focused initiatives ahead of a meeting with top tech CEOs from Alphabet, Anthropic, OpenAI, and Microsoft. Key actions include allocating $140 million in funding to seven new National AI Research Institutes and holding an independent exercise with major tech players to assess generative AI systems' adherence to the Biden administration's AI Bill of Rights blueprint. Continue Reading →

For reasons that are theorized – but not fully understood – human beings "default-to-trust." This strategy has stood the test of time. We generally trust others until given a reason to do otherwise. However, as the output from large language models and generative AI becomes increasingly indistinguishable from that produced by humans, it's time to consider a new paradigm: "default to distrust." Continue Reading →

Chat interfaces for GPT-4 may soon take a backseat to a new approach: autonomous agents. Emerging versions such as AutoGPT, BabyAGI, and the no-code GodMode promise a simplified user experience by focusing on achieving user goals with minimal input. Although it's still early days, autonomous agents are poised to revolutionize how we interact with technology and each other. Continue Reading →

The human creative community, the U.S. Copyright Office, and AI-enabled creators all have different ideas about where copyrightable human creativity begins. Are you entitled to royalties if your work is used to help train a model? If a model is prompted to create something that is inspired by a bunch of different styles is that work "original?" If a human uses what they've learned to craft a prompt is the result of that prompt "original?" Humans learn by mimicking. In fact, human beings are the most successful mimics on the planet. “Watch Mommy!” or “Daddy will show you how to do it,” or “Repeat after me” are phrases we have heard all of our lives. You will learn about this subject by mimicking the words, acronyms and concepts written here – that’s why you are reading. Generative AI learns in a very similar way. Has it crossed a line? If it has, we’ve been walking up to the edge for decades. Let's explore. Continue Reading →

Be honest – how often do you read the Terms & Conditions for the online services you use? If you're like most people, the answer is "almost never." That's fine for some things, but today, I want you to read Section 3, Paragraph (c) of OpenAI's Terms of Use. It is the most important thing you are going to read this morning. Continue Reading →

The rise of generative AI has unlocked the potential for exponential productivity increases for everyone. We're seeing 5 to 15% productivity increases from workers who are simply leveraging ChatGPT. Imagine if you could create a custom AI that was an expert in your industry, an expert in the inner workings of your company, and always up to date. The term for processing your data and then using it to enhance a GPT model's expertise for your specific purposes is "embedding." There are a couple of ways to accomplish this, but one stands out as more practical than the other. Let's explore. Continue Reading →

This morning, I did something I rarely do: I signed an open letter alongside Elon Musk, Steve Wozniak, Yuval Noah Harari, and a group of AI experts and tech executives, urging top AI researchers to pause development of AI systems more advanced than GPT-4 for six months. Continue Reading →

This essay was originally published on November 16, 2019. It was updated and republished on November 11, 2022. Adobe announced Firefly, its generative AI assistant, on March 21, 2023. If you want a clear example of the exponential pace of technological change, most of the capabilities I hypothesized (and many I did not think of) are features of Firefly. Continue Reading →

The next generation of the World Wide Web (web3) is being built on blockchains using tools and technologies similar to the ones that empower cryptocurrency, NFTs, smart contracts, and the world of DeFi. While there is no agreed-upon technical definition for web3, my favorite philosophical definition is: "A decentralized platform that empowers both users and creators to share in the value they create." It's a nice idea, but how would it actually work? Continue Reading →