Terra Terwilliger

At Joint California Summit on Gen AI today hosted by Stanford and UC Berkeley, Kapor Center, San Francisco Foundation, CA Governors Office and others w/ top leaders from policy, research, philanthropy and community. Gavin Newsom brilliantly moderated a conversation with Fei-Fei Li and Jennifer Chayes on the risks and opportunities associated with #genai. It was very thought provoking--encourage you to check it out. After, I asked the new ChatGPT-4 to outline key opportunities and challenges associated with #ai in the ed sector. Here is what the advanced version produced: As we stand on the brink of an AI revolution, the education sector faces unprecedented opportunities and significant challenges. Opportunities: Personalized Learning: AI-driven platforms can tailor educational experiences to individual student needs, making learning more effective and engaging. Imagine a classroom where each student has a personalized tutor, adapting lessons in real-time based on their progress! Efficiency: AI can automate administrative tasks, allowing educators to focus more on teaching and student interaction. From grading assignments to managing schedules, AI can streamline operations and reduce workloads. Access to Education: AI can democratize education by providing high-quality resources to students in remote or underserved areas. Through online courses and virtual classrooms, we can bridge the gap and offer equal learning opportunities to all. Enhanced Engagement: Interactive AI tools, such as chatbots and virtual reality, can make learning more interactive and enjoyable. These technologies can bring subjects to life, sparking curiosity and a love for learning. Challenges: Data Privacy: With the increased use of AI, the collection and storage of student data raise significant privacy concerns. Ensuring robust data protection measures is crucial to maintaining trust and security. Equity Issues: While AI has the potential to provide access, it can also widen the gap if not implemented thoughtfully. Ensuring that all students have access to the necessary technology and resources is essential to prevent further disparities. Teacher Training: Educators need to be adequately trained to integrate AI tools into their teaching methods. Professional development and ongoing support are vital to harnessing the full potential of AI in education. Ethical Considerations: The use of AI in education must be guided by ethical principles. This includes addressing biases in AI algorithms and ensuring that the technology is used to support, not replace, the human elements of teaching and learning. As we navigate this transformative era, collaboration between educators, technologists, policymakers, and communities will be key to leveraging AI for the benefit of all learners. What are your thoughts on the role of AI in education? Share your insights and let's continue this important conversation! 🤖📚✨ #AIinEducation #EdTech #FutureOfLearning #InnovationInEducation

18

3 Comments

From my students at Cornell Tech and Massachusetts Institute of Technology, to our work at Startups & Society Initiative and Responsible Innovation Labs, continue to see the brightest young people hungry for models of business and technology that reject 'growth at all costs' and embrace more deliberate notions of value creation. From climate, to democracy, to public health, the next generation holds deep intuition that value is not captured by 'valuation.' Its not that they outright don't believe in profit, but they know it's essential to build with intention and responsibility, mindful of broader societal impacts. It was a treat to join Keshav Rastogi — a young person already building and leading with his values — for his Podcast "Playgrounds to Pixels." Keshav asks thoughtful questions about the business motivations for responsible tech, my work at MEET - Middle East Entrepreneurs of Tomorrow, and how young people can start thinking about responsible tech. Thanks again for having me, Keshav.

52

5 Comments

Over the past couple of years, there's been a noticeable decrease in the openness to share cutting-edge LLM research. I've been fortunate to have lead a group working on advanced Generative AI challenges and to have the opportunity to share our work publicly, potentially impacting a broader audience. Exciting news! Our research, centered on enhancing batch size scalability for LLMs, has been accepted at ICML 2024. We addressed the vital challenge of single-context batch sampling, crucial for generating multiple completions from one context, a common scenario in applications like code-editing IDE tools or ranking among multiple generations. Incremental decoding in such scenarios presents a memory IO-intensive challenge, particularly for large batches and context lengths. Our efforts improve efficiency, resulting in reduced latency, making it suitable for real-time applications such as massively-parallel answer generation without a significant increase in latency. Additionally, it enhances performance when paired with postprocessing techniques like reranking, contributing to overcoming the challenge of scaling model size for improved performance with reranking. Our solution introduces context-aware bifurcated attention, dividing attention in the generalized multi-query family into context and decoding components during incremental decoding. This bifurcation maintains the same computational complexity and delivers identical results as the original attention mechanism, while notably reducing memory IO costs and associated latencies in scenarios with large batch sizes and context lengths. This approach enables generating multiple real-time completions without incurring significant additional latency costs or allows for much larger batch sizes, thereby enhancing ranking performance. For example, using a 16B multi-head model with a 2k context length, we increased the batch size to 128 with bifurcated attention, compared to a batch size of only 5 without it, resulting in pass@k increasing from 59.0% to 84.6% Paper: https://lnkd.in/gXbhc4jT Kudos to the great team I had opportunity to work with Ben Athiwaratkun Sanjay Krishna Gouda Haifeng Qian qing sun Hantian Ding Ramesh Nallapati Sujan Gonugondla Jun Wang Sudipta Sengupta Bing Xiang

128

2 Comments

Look forward to speaking at the Enterprise Generative AI Summit this week along side esteemed panelists Dr. Yvonne Lutsch and Mike Shirazi on the importance of data relevancy and data privacy considerations in next wave of LLM investments. #dataprivacy #GenAI #LLM https://lnkd.in/efj2egAG

82

Science-fiction is inherently speculative. Author George #Orwell of ‘#1984’ predicted concepts such as doublethink, little earbuds and society’s obsession with the black mirror screens that dominate our lives — and according to the BBC, novelist and short story author JG Ballard, of “High-Rise” and “Crash” fame, has predicted the current tool and cultural fixation of #GenerativeAI in the modern society — only he didn’t just speculate them within a story, he is actually one of the first use-cases of a generative AI used in a creative context — Ballard was so fascinated with the growing technological advancement of computers that he used one to compose poems, essentially the precursor to what we have now with ChatGPT. Today, the debate about the creative use of generative AI and its value on creative industries is an extremely fraught and controversial topic, but if controlled in a space that doesn’t replace human labor or try to equate itself to traditional art forms, it is interesting to see what computers can do and how they are creatively limited in comparison. Ballard’s two poems, The Yellow Back Novels and Machine Gun City are generated, and both include notes on how Ballard achieved producing them. Regardless of quality (Ballard himself probably wouldn’t describe these two poems as his best work), it is interesting to see primitive versions of the technology that we utilize on the daily today, and how these moral conversations that concern them would continuously evolve.

1

1 Comment

Kudos to the Massachusetts Institute of Technology for convening one of the largest interdisciplinary gatherings of #AI leaders, entrepreneurs, researchers and academics. Thank you to John Werner, Sandy Pentland and Randall Lane for bringing us together. Together with fellow panelists Jeffrey Saviano, Dr. Sasha Luccioni and Ra'ad Siraj we excavated AI Governance, Sustainability, and Scale, and the imperatives of ethics, corporate values and responsible business with a principled-based approach to #AI. As with driving, like business, brakes let you go faster. It feels counter-intuitive, but without brakes we would drive really slowly. A great moment for advancing #AI, with the MIT Media Lab hosting 2000+ attendees, 180 speakers, 80 AI startups, 67 "lightning talks", 36 panels, and 35 technical talks, spread over 4 tracks. #artificalintelligence #startups #entrepreneurship #mit #tech #esg #sustainability #LLM #generativeAI

138

5 Comments

Interesting and compelling speech about the current state of global political and economical systems, that we have created for our future generations and now asking them to fix it.

1

I just published a new post "Innovating with conviction: lessons from Marc Pincus (Zynga founder)" https://lnkd.in/epTFRxWD We learn an important innovation framework from Marc, including: - The goal of innovation: getting to true signal - Develop your core idea thesis, and create multiple idea variants. Keep your idea variants separate from your idea thesis. - Cycle through your idea variants quickly to take the maximum number of shots on goal - Get to minimum viable idea state, not even minimum viable product, to test your idea variant Lots of practical innovation guidance from Marc Pincus, I've learned a lot and I'm immediately putting his framework into practice. I encourage anyone working on 0 -> 1 projects to do the same!

16

1 Comment

Do you want to learn the 'secrets' of AI and ML? [Chuckes facetiously.] Might one recommend stopping and reading a few books? Why? Well, it pays to train one's (own) neural network. Joking aside, it is a fascinating field, and I readily admit I enjoy learning about it. Note: I do not consider myself an expert in anything; these are simply books I like. You may, too. Artificial Intelligence: A Modern Approach https://lnkd.in/d-NM8hWe Deep Learning https://lnkd.in/duHQaUSy Approaching (Almost) Any Machine Learning Problem Alas, it's out of print, but a great read, written by a fine fellow. 🚀 Abhishek Thakur Mathematics for Machine Learning https://lnkd.in/dCvfcqHU beyond the above, I highly recommend reading more broadly: Travels with Herodotus by Ryszard Kapuscinski https://lnkd.in/dNafjjiF

25

7 Comments

Excerpt from an Angel Squad memo that we invested in recently: “[Company] is led by a team of scientists and engineers with PhD-level expertise in Machine Learning, Engineering Simulations, and Computational Materials Science. The team is headed by CEO, [Name], who previously led AI development at two early-stage startups. [Name] is joined by CTO, [Name], who has deep experience in machine learning and materials science, and Chief Science Officer, [Name], a doctoral researcher at the [School]. “ Investing at the earliest stages of value creation often means taking big bets on a founding team. At Angel Squad and Hustle Fund, we love founders with deep expertise and super domain knowledge, particularly in deep tech verticals.

28

3 Comments

📣 Calling all the #emergingtech #founders and #builders out there. Do you ever worry that when the #regulators knock, they'll completely misunderstand what you're building? Do you feel like the headlines about your industry is more gossip than tech, but you're the one paying the price? Let's #disrupt this! 🚀 I've been working tirelessly to create a summit that will train you so you can lead the narrative, and stop playing defense. Join us at University of California, Berkeley on May 13th for TECHVocacy 2024: Emerging Tech Advocacy Summit! What makes this summit so different? No long winded panels to talk about the problem or to talk about the tech (#ai #crypto #web3 #av #socialmedia). We have 🔵 Experts to deliver lightning talks that will SHOW (not tell) how to talk about your emerging and disruptive tech with policymakers/regulators 🔵 Real-time media training demos Co-hosted by Berkeley Center for Law and Business Berkeley Haas Entrepreneurship Berkeley Center for Responsible Decentralized Intelligence with support from Berkeley Space Center. Admission is free, but registration is limited! Sign up soon!

32

4 Comments

Watched a fascinating fireside chat between Patrick Collison and Jensen Huang at Stripe Sessions. Worth sharing my notes... Jensen started Nvidia when he was 29 and has been CEO for 31 years. He has 60 direct reports and does not have 1:1s. Issues are discussed in bi-weekly staff meetings. Feedback is delivered real-time, in public. "Why should the person who made a mistake be the only one to learn from that mistake?" "There's value in learning through other people's tragedies, that's why they use case studies in school." 😂 "I'd rather torture you into greatness, because I believe in you." So he rarely fires people...instead he'll keep torturing you. 😂 On work/life balance. He "works" 24/7. Work to him is fantasizing and dreaming about "what could be?" He avoids regular operational meetings. He only focuses on problem / idea / brainstorming meetings. He loves to focus on zero billion markets. To create something that's never been done, that's hard, that no one focuses on. Doesn't like the concept of "taking share"...prefers creating something entirely new. Nvidia invented a concept called accelerated computing (CUDA), that didn't sell for 10 years. But his core beliefs suggested customers just hadn't caught up yet. His beliefs eventually paid off...none of their AI services would be possible without accelerated computing. "First principle" thinking was what kept him and Nvidia focused. They got creative in finding applications for CUDA to sustain them "just enough" until the rest of the market caught up…seismology, biotechnology, robotics. The world has installed $1T worth of data centers for general purpose computing. In 4-6 years, the installed base of servers will need to be replaced with accelerated computing devices, Nvidia devices. Data centers will turn into AI factories that take in energy and produce tokens that can be used to produce language, images, videos, chemical formulas, robotic signals, etc. Electricity markets took energy and turned it into electricity sold at $/Mwh. AI will take electrons and turn them into AI tokens in the same way. Error rates on AI generated content are declining 50% every 6 months. You won't lose your job due to AI, you'll lose it because another candidate used AI better than you. Your company won't go out of business because of AI, it will go out of business because a competitor used it more effectively. Nvidia has 28K employees and a market cap approaching $2T. Compared to Apple and Microsoft who both have 4-6X more employees. Jensen uses words like “love and care” as part of his business ethos to encourage an appreciation for craftsmanship. Customers don't ask for love and care and craftsmanship, but when they experience it in a product or service, they appreciate it and value it. Unconventional, original, refreshing thinking from someone that history will rate as a candidate as one of the business world’s GOATs.

14

5 Comments

I like this idea of "technical taste". It gets into where software engineering is as much creative art as it is science. Four great takeaways (IMHO) for software devs in this era in particular: 1. Aspiring engineers should cultivate a sense of curiosity, experiment with different tools and technologies, and embrace a mindset of continuous learning. 2. AI has the potential to streamline processes and enhance productivity for engineers, but it may also lead to disruptions in traditional software development workflows. 3. Developing technical taste and judgment is essential for making informed decisions about which technologies and approaches to pursue. 4. Collaboration and open-mindedness are key to leveraging the full potential of AI and staying ahead of technological advancements. I also really liked these two quotes from Sam Schillace: 1. "The right time to do something is when you have that feeling in the pit of your stomach that's like, 'oh, this is a great idea and it's going to suck to build because nothing's ready yet.'" 2. "Technical taste is like, 'how well have you consolidated that set of experiences and heuristics into judgment that you can apply accurately when you see new things?'" 3. "It may be the case that very small teams can do very large projects, or like we were talking about before, it may be the case we're just going to get really ambitious about what we try to do with the same size of teams, which is kind of where I would put my money." Great job on continuing to put out great episodes, Brett Berson and team First Round Capital! #ai #engineering #softwaredevelopment

5

Earlier this week Texture, led by CEO & co-founder Sanjiv Sanghavi, announced its $7.5m seed round of fundraising with participation from Abstract Ventures, Day One Ventures, Equal Ventures, Lerer Hippeau, and a handful of angels. Dorothy Chang and I feel fortunate to count ourselves among the “handful of angels.” We didn’t invest just because Sanjiv is a close friend - though knowing him for over a decade has given us a lot of insight into his mind and how he operates. We invested for a lot of the typical reasons a VC would: Solving big problems with big potential returns - Venture investments only make sense if they have the chance to make massive returns. Texture is building the data infrastructure for energy networks - becoming a common data collection and sharing platform for every player in the energy grid, no matter the type of energy. There is already a huge gaping problem and a huge market here. Building for an inevitable future - On a macro level, as a society, we are moving more and more toward complex, interconnected energy sources as we optimize for the right balance of efficiency, sustainability and costs. Unique insight and founder-market fit - At a first glance, Texture seems like a big departure from what we originally knew Sanjiv from, as a Co-Founder of ClassPass. But since then, he has also played roles as the Chief Product officer at Arcadia, a successful energy data company (Arcadia’s CEO is also an investor in Texture) as well as an investor at Day One Ventures, investing in the energy/climate tech sector.  More details in this week's post: https://lnkd.in/eGBzXHwC

44

4 Comments

I’m thoughtful about the impact that the dissolution of OpenAI’s risk and ethics team led by Ilya Sutskever and Jan Leike might have on the direction of the company’s tech & product development. And - this strengthens my resolve to earn a Masters of Data Science & AI degree over the next 18 months, weighted toward the deeply important conversation around building tech within a framework of ethics, equity and risk mitigation. We must find an equilibrium that balances the raw speed of AI innovation with intentionality. Can’t wait to get started! OpenAI #ai #BuildWithIntention #venturestudios #vc

16

2 Comments

One of the oddest and most dangerous ‘dividends’ arising from the hypergrowth of Artificial Intelligence is the rise of what policy expert Jamie Neikrie and I call ‘Artificial History.’ We wrote up a piece for Tech Policy Press about the ability for generative AI to create highly credible but entirely fictional synthetic history. Now, bad actors have the ability to arm their claims with retroactive points of evidence spun out of fake first-hand accounts, on-the-ground testimony, or deepfake trusted messengers. If you can rewrite the narrative of where we come from, you can make a different argument about where we should be going. See more here: https://lnkd.in/eqxTQvae

18

10 Comments