The AI talents wars are just getting started

The AI Talent Wars Are Just Getting Started by Alex Heath • The Verge - Artificial Intelligence 20 déc. 2024, 21:26 For my last issue of the year, I’m focusing on the AI talent war, a theme I’ve bee

Here's the reformatted content for the Socra: # The AI Talent Wars Are Just Getting Started by Alex Heath • The Verge - Artificial Intelligence December 20, 2024, 21:26 For my last issue of the year, I’m focusing on the AI talent war, a theme I’ve been covering since this newsletter launched almost two years ago. And keep reading for the latest from inside Google and Meta this week. ### “It’s Like Looking for LeBron James” This week, Databricks announced the largest known funding round for any private tech company in history. The AI enterprise firm is in the final stretch of raising $10 billion, almost all of which is going to buy back vested employee stock. How companies approach compensation is often undercovered in the tech industry, even though these strategies play a crucial role in determining which company gets ahead faster. Nowhere is this dynamic more intense than in the war for AI talent. To better understand what’s driving the state of play going into 2025, this week I spoke with **Naveen Rao**, VP of AI at Databricks. Rao is one of my favorite people to talk to about the AI industry. He’s deeply technical but also business-minded, having successfully sold multiple startups. His last company, MosaicML, sold to Databricks for $1.3 billion in 2023. Now, he oversees the AI products for Databricks and is closely involved with its recruiting efforts for top talent. **Why is this round mostly to help employees sell stock? Because $10 billion is a lot. You can do a lot with that.** The company is a little over 11 years old. There have been employees that have been here for a long time. This is a way to get them liquidity. Most people don’t understand that this is not going into the balance sheet of Databricks. This is largely to provide liquidity for past employees and for current and new employees. It ends up being neutral on dilution because they’re shares that already exist. They’ve been allocated to employees, and this allows them to sell those to cover the tax associated with those shares. **How much of the rapid increases in AI company valuations have to do with the talent war?** It’s real. The key thing here is that it’s not just pure AI talent, people who come up with the next big thing, the next big paper. We are definitely trying to hire those people. There is an entire infrastructure of software and cloud that needs to be built to support those things. When you build a model and want to scale it, that actually is not AI talent, per se. It’s infrastructure talent. The perceived bubble around AI has created an environment where all of those talents are getting recruited heavily. We need to stay competitive. **Who is being the most aggressive with setting market rates for AI talent?** OpenAI is certainly there. Anthropic. Amazon. Google. Meta. xAI. Microsoft. We’re in constant competition with all of these companies. **Would you put the number of researchers who can build a new frontier model under 1,000?** Yeah. That’s why the talent war is so hot. The leverage that a researcher has in an organization is unprecedented. One researcher’s ideas can completely change the product. That’s kind of new. In semiconductors, people who came up with a new transistor architecture had that kind of leverage. That’s why these researchers are so sought after. Somebody who comes up with the next big idea and the next big unlock can have a massive influence on a company’s ability to win. **Do you see that talent pool expanding in the near future or is it going to stay constrained?** I see some aspects of the pool expanding. Being able to build the appropriate infrastructure and manage it, those roles are expanding. The top-tier researcher side is the hard part. It’s like looking for **LeBron James**. There are just not very many humans who are capable of that. I would say the [Inflection-style acquisitions](https://www.theverge.com/2024/7/1/24190060/amazon-adept-ai-acquisition-playbook-microsoft-inflection) were largely driven by this mentality. You have these concentrations of top-tier talent in these startups, and it sounds ridiculous how much people pay. But it’s not ridiculous. I think that’s why you see [Google hiring back Noam Shazeer](https://www.theverge.com/2024/8/2/24212348/google-hires-character-ai-noam-shazeer). A guy we had at my previous company that I started, Nervana, is arguably the best GPU programmer in the world. He’s at OpenAI now. Every inference that happens on an OpenAI model is running through his code. You start computing the downstream cost, and it’s like, “Holy shit, this one guy saved us $4 billion.” **What’s the edge you have when you’re trying to hire a researcher at Databricks?** You start to see some selection bias among different candidates. Some are AGI or bust, and that’s okay. It’s a great motivation for some of the smartest people out there. We think we’re going to get to AGI through building products. When people use technology, it gets better. That’s part of our pitch. AI is in a massive growth phase but it’s also hit peak hype and is on the way down the [Gartner hype curve](https://en.wikipedia.org/wiki/Gartner_hype_cycle). I think we’re on that downward slope right now, whereas Databricks has established a very strong business. That’s very attractive to some because I don’t think we’re so susceptible to the hype. **Do the researchers you talk to really believe that AGI is right around the corner? Is there any consensus on when it’s coming?** Honestly, there’s not a great consensus. I’ve been in this field for a very long time and I’ve been pretty vocal in saying that it’s not right around the corner. The large language model is a great piece of technology. It has massive amounts of economic uplift and efficiencies that can be gained by building great products around it. But it’s not the spirit of what we used to call AGI, which was human or even animal-like intelligence. These things are not creating magical intelligence. They’re able to slice up the space that we’re calling facts and patterns more easily. It’s not the same as building a causal learner. They don’t really understand how the world works. You may have [seen Ilya Sutskever’s talk](https://www.theverge.com/2024/12/13/24320811/what-ilya-sutskever-sees-openai-model-data-training). We’re all kind of groping in the dark. Scaling was a big unlock. It was natural for a lot of people to feel enthusiastic about that. It turns out that we weren’t solving the right problem. **Is the new idea that’s going to get to AGI the test-time compute or “reasoning” approach?** No. I think it’s going to be important for performance. We can improve the quality of answers, probably reduce the probability of hallucinations, and increase the probability of having responses that are grounded in fact. It’s definitely a positive for the field. But is it going to solve the fundamental problem of the spirit of AGI? I don’t believe so. I’m happy to be wrong, too. **Do you agree with the sentiment that there’s a lot of room to build more good products with existing models, since they are so capable but still constrained by compute and access?** Yeah. Meta started years later than OpenAI and Anthropic, and they basically caught up, and xAI caught up extremely fast. I think it’s because the rate of improvement has essentially stopped. Nilay Patel compares the AI model race to early Bluetooth. Everyone keeps saying there’s a fancier Bluetooth but my phone still won’t connect. You see this with every product cycle. The first few versions of the iPhone were drastically better than the previous versions. Now, I can’t tell the difference between a three-year-old phone and a new phone. I think that’s what we see here. How we utilize these LLMs and the distribution that has been built into them to solve business problems is the next frontier. --- ## AI News Analysis Framework ### 1. Summary This article delves into the escalating competition for AI talent, exemplified by Databricks' unprecedented $10 billion funding round. Through an interview with Naveen Rao, VP of AI at Databricks, it provides insights into the current landscape of AI talent acquisition, the dynamics of company valuations driven by this competition, and broader industry perspectives on the pursuit of AGI. ### 2. Big Ideas * **Elite AI Talent Scarcity:** * Fewer than 1,000 researchers globally are capable of building frontier models. * Individual researchers can generate billions in value through strategic optimizations. * The search for such talent is likened to "looking for LeBron James," emphasizing its extreme rarity. * **Evolving Talent Acquisition Strategies:** * There's a multi-layered demand for talent, encompassing both pure AI researchers and critical infrastructure experts. * Compensation strategies, including stock buybacks and equity, are key drivers of company valuations. * Strategic acquisitions, often focused on securing top talent, are becoming a common tactic. * **AGI Development Reality Check:** * Current Large Language Models (LLMs) are distinct from Artificial General Intelligence (AGI). * The AI field is transitioning from a primary growth phase to an optimization phase. * The focus is shifting from fundamental model development to practical applications and product building. ### 3. Knowledge Gaps & Concepts #### Key Technical Concepts: * **Frontier Models:** The latest generation of large AI models that push technological boundaries and define the state of the art. * **Test-time Compute:** The computing resources utilized during the inference phase of a model, where it processes new data to make predictions or generate outputs. * **Causal Learning:** An advanced form of AI that seeks to understand cause-effect relationships, distinguishing it from systems that merely identify patterns or correlations. #### Business Concepts: * **Liquidation Preference:** A clause in venture capital agreements that determines the order and amount of payout to investors and employees upon a liquidity event (e.g., acquisition or IPO). In this context, it refers to how employee stock options provide liquidity. * **Talent Leverage:** The disproportionate impact a highly skilled individual, particularly in a rare field like advanced AI research, can have on a company's product, efficiency, and overall value. * **Gartner Hype Cycle:** A graphical representation of the maturity, adoption, and social application of specific technologies, often showing an initial "peak of inflated expectations" followed by a "trough of disillusionment." ### 4. Extra Context & Foundational Knowledge #### Talent Market Dynamics: * **Major Players:** Key competitors in the AI talent war include OpenAI, Anthropic, Amazon, Google, Meta, xAI, and Microsoft. * **Compensation Structures:** AI talent compensation typically includes a significant mix of salary and equity components, designed to attract and retain top researchers. * **Infrastructure Importance:** Beyond pure AI research talent, the ability to build and manage robust AI infrastructure is becoming equally, if not more, crucial for scaling and deploying models. #### Industry Evolution Indicators: * **Slowing Model Improvements:** Naveen Rao suggests a slowdown in the fundamental rate of improvement for frontier models. * **Shift to Applications:** The industry is pivoting toward leveraging existing powerful models to solve real-world business problems and optimize performance. * **Business Model Sustainability:** There's a growing emphasis on building sustainable business models around AI, moving beyond the initial hype toward practical value creation. ### 5. Expert Perspective Access To gain further expert insights into the AI talent war and industry trends, we recommend: 1. **Following Technical Leaders:** Pay attention to the public statements, research papers, and strategic moves of influential figures like Ilya Sutskever (OpenAI), Noam Shazeer (Google, Character AI), and Naveen Rao (Databricks). 2. **Tracking Company Strategies:** Monitor funding rounds, strategic partnerships, and acquisitions made by major AI companies, as these often reveal their talent priorities and growth directions. 3. **Studying Implementation Success Stories:** Analyze how different companies successfully implement and derive value from AI, moving beyond theoretical advancements to practical, impactful applications. ### 6. Knowledge Progression Path Here’s a structured path to deepen understanding of the AI talent landscape: #### Level 1: Understanding the Basics * Identify the different types of AI talent roles (e.g., researcher, engineer, data scientist). * Understand the basic components of AI infrastructure (e.g., GPUs, cloud platforms, data pipelines). * Grasp fundamental concepts of company valuation in the tech industry. #### Level 2: Technical Depth * Delve into specific AI model architectures and scaling techniques. * Explore advanced methods for infrastructure optimization and cost reduction. * Learn about key performance metrics and evaluation methodologies for AI models. #### Level 3: Strategic Insight * Analyze the long-term dynamics of the AI talent market and its impact on innovation. * Understand various company positioning strategies in the competitive AI landscape. * Predict future evolution patterns of the AI industry and their implications for business and careers. ### 7. Follow-up Questions & Ideas #### Technical Questions: 1. What specific skills differentiate top-tier AI researchers, and how can these be cultivated? 2. How is the demand for and skill set of AI infrastructure talent evolving alongside AI model advancements? 3. What are the quantitative and qualitative metrics used to define and assess "frontier model" capability? #### Business Strategy: 1. Are current AI talent compensation levels sustainable in the long term, or will they stabilize/decline? 2. What alternative talent development and retention strategies are companies exploring beyond direct hiring and acquisitions? 3. How might market consolidation among major AI players affect the distribution and availability of top AI talent? #### Future Implications: 1. What impact will the AI talent war have on AI education and training programs globally? 2. How will the roles and responsibilities of AI researchers evolve as the field shifts from fundamental breakthroughs to optimization and application? 3. What changes might we see in the patterns of company formation and funding in the AI sector due to talent scarcity? #### Personal Development: 1. What are the key skills needed for different AI career paths (e.g., research, engineering, product management)? 2. How can individuals best understand and maximize their value creation in various AI roles? 3. What strategies can professionals use to effectively position themselves in the highly competitive AI talent marketplace? --- ### What specific skills differentiate top-tier AI researchers? Top-tier AI researchers distinguish themselves through a combination of advanced technical expertise, innovative thinking, interdisciplinary knowledge, and a strong ethical foundation. As the field of artificial intelligence (AI) continues to evolve rapidly, the skills required to excel at the highest levels are becoming increasingly diverse and specialized. #### 1. Advanced Mathematical and Statistical Knowledge Mathematics is the backbone of AI. Top researchers must have a deep understanding of linear algebra, calculus, probability, and statistics to design and analyze algorithms effectively. Key areas include optimization techniques for training models, probabilistic models and Bayesian inference, and advanced topics like differential geometry for neural networks and tensor calculus for deep learning architectures. These skills are critical for developing new algorithms, improving model efficiency, and solving complex problems in areas like generative AI, reinforcement learning, and multimodal systems. #### 2. Mastery of Machine Learning and Deep Learning AI research is heavily focused on machine learning (ML) and deep learning (DL). Researchers must not only understand existing methods but also innovate and push the boundaries of these technologies. This includes advanced ML techniques (e.g., transfer learning, meta-learning, and federated learning), deep learning architectures like transformers, GANs (Generative Adversarial Networks), and diffusion models, as well as reinforcement learning advancements. These skills enable researchers to tackle cutting-edge challenges in natural language processing (NLP), computer vision, robotics, and autonomous systems. #### 3. Programming and Technical Proficiency The ability to implement ideas into code is essential for experimentation and validation. Proficiency in Python, R, or Julia for algorithm development is crucial, alongside expertise in ML frameworks like TensorFlow, PyTorch, and JAX. Familiarity with big data tools (e.g., Apache Spark, Hadoop) and cloud platforms (e.g., AWS, Google Cloud, Azure) is also vital. Efficient coding and debugging skills allow researchers to prototype, test, and optimize AI models quickly. #### 4. Research and Analytical Skills AI research is inherently experimental. Researchers must design experiments, test hypotheses, and analyze results rigorously. This involves designing reproducible experiments, benchmarking models, and analyzing large datasets to extract meaningful insights. Staying updated with the latest research papers and trends in AI is also paramount. These skills are crucial for publishing high-impact research and contributing to the scientific community. #### 5. Creative Problem-Solving and Critical Thinking AI research often involves tackling novel and complex problems. Creativity and critical thinking are essential for developing innovative solutions. This includes breaking down complex problems into manageable components, questioning assumptions and exploring unconventional approaches, and balancing open-mindedness with skepticism to validate findings. These skills drive breakthroughs in areas like generative AI, quantum computing, and AI interpretability. #### 6. Interdisciplinary Knowledge AI is increasingly applied across diverse fields, from healthcare to finance to climate science. Understanding domain-specific challenges enhances the relevance and impact of AI solutions. Knowledge of biology for AI in drug discovery, understanding financial systems for AI in fintech, or familiarity with physics for AI in robotics and autonomous systems are examples. Interdisciplinary expertise allows researchers to tailor AI solutions to specific industries and societal needs. #### 7. Ethical Judgment and Social Responsibility As AI systems become more pervasive, ethical considerations are critical to ensure fairness, transparency, and accountability. Key skills include understanding AI ethics frameworks (e.g., IEEE, EU guidelines), identifying and mitigating biases in datasets and models, and anticipating the societal impacts of AI technologies. Ethical judgment ensures that AI advancements align with societal values and contribute positively to humanity. #### 8. Advanced Model Interpretability and Explainability Trustworthy AI requires models that are interpretable and explainable, especially in high-stakes applications like healthcare and law. Expertise in interpretability techniques like SHAP, LIME, and Integrated Gradients, and developing methods to visualize and explain model decisions, are essential for building AI systems that stakeholders can trust and understand. #### 9. Quantum Computing and Emerging Technologies Quantum computing is poised to revolutionize AI by enabling faster and more efficient computations. Understanding quantum mechanics and quantum algorithms, along with familiarity with platforms like IBM Quantum Experience and Microsoft Quantum Development Kit, can accelerate AI research in optimization, cryptography, and large-scale simulations. #### 10. Collaboration and Communication Skills AI research often involves interdisciplinary teams. Effective communication is crucial for collaboration and knowledge sharing. This means writing clear and concise research papers, presenting findings to both technical and non-technical audiences, and collaborating with engineers, domain experts, and policymakers. These skills foster teamwork and ensure that research findings are accessible and impactful. #### 11. Proficiency in Data Engineering and Big Data Analytics AI models rely on large datasets. Managing and processing these datasets efficiently is a critical skill. This includes data preprocessing, cleaning, and augmentation, using big data platforms like Apache Hadoop and Spark, and extracting insights from unstructured data (e.g., text, images, videos). These skills are essential for training robust and scalable AI models. #### 12. Lifelong Learning and Adaptability The AI field evolves rapidly, with new tools, techniques, and challenges emerging constantly. Top researchers continuously update their knowledge with the latest research and technological advancements, adapt to new tools and frameworks, and cultivate a growth mindset. Lifelong learning ensures that researchers remain at the forefront of AI innovation. #### 13. Engagement with the AI Community Collaboration and knowledge sharing within the AI community drive innovation and progress. Active participation in conferences, workshops, and hackathons, contributing to open-source projects, and engaging with professional networks and forums foster collaboration, mentorship, and access to cutting-edge resources. #### 14. Focus on Foundational Principles Over Tools Tools and frameworks evolve quickly, but foundational principles remain constant. Top researchers understand core AI concepts like optimization, generalization, and overfitting, building transferable knowledge that applies across various tools and technologies. This approach ensures adaptability and long-term success in the AI field. **Conclusion:** Top-tier AI researchers excel by combining deep technical expertise with creativity, ethical awareness, and interdisciplinary knowledge. They are not only skilled in the latest tools and techniques but also possess a strong foundation in mathematics, programming, and research methodology. Their ability to think critically, collaborate effectively, and adapt to new challenges sets them apart as leaders in the ever-evolving field of AI. By mastering these skills, researchers can drive innovation, address societal challenges, and shape the future of artificial intelligence.

By Romain Peter