Welcome back to the AI Geekly, by Brodie Woods. Your curated 5-minute read on the latest developments in the rapidly evolving world of AI.
Cutting through clickbait, hype, and old media, the AI Geekly provides the perspective of an AI strategist and capital markets veteran, informed by nearly two decades on the front lines.
As we enter the Age of AI, staying informed and critical is key.
AI Quote of the Week
“The shift to AI… will be the most profound shift we’ll see in our lifetimes. It will touch every sector, every industry, every business function. And significantly change the way we live and work. This isn’t just the future. We’re already starting to experience the benefits right now."
-Sundar Pichai, CEO Alphabet + Google at this week’s Google Cloud Next ‘23 Conference Keynote
TL;DR - Exec Summary
Google’s Annual Cloud AI Conference, Censors Break AI Models, Incumbent Upstarts Nip at Nvidia’s Heels, AI21 Valued at $1.4 Bn, Vendors Commit to “Your Data is YOUR Data”
Google made several solid announcements at its annual Google Cloud Next (GCN) conference this week, expanding the capabilities of its AI cloud offerings for developers and users with practical tools to enhance productivity available today. Researchers demonstrated that alignment negatively impacts model performance, raising important considerations around censorship and authority. Turning to the topic of silicon (chips), we see very early potential signs of disruption to Nvidia’s uncontested AI chip monopoly beginning to emerge on a few fronts: AMD’s latest ROCm releases have enabled AMD GPUs to run modern GenAI models as a drop-in replacement for Nvidia CUDA; Intel’s Habana GAUDI 2’s built-in memory-loading hardware accelerator de-bottlenecks certain AI workloads driving outperformance vs Nvidia’s H100s on some tasks by > 2x. Tel-Aviv-based AI21 Labs is the latest AI company to see a large capital infusion and raising $155 mm in a Series C financing valuing the company's pay-as-you-go platform for custom text-based business apps at $1.4Bn. Summing-up several days of conference presentations by Google, and its partners the prevailing message was around usability and security –these solutions are ready for prime-time and vendors are willing to do what is necessary to ensure they meet enterprise needs around data security (i.e. your data doesn’t go into training their AI models).
With its announcements this week, Google Cloud has demonstrated their ability to read the room and, at first glance, is delivering a suite of tools that can add benefit across the infra, development, and user/customer value chain. We’re eager to see how AWS, Microsoft, and Meta respond, given each company’s relative position in the GenAI landscape.
-Read on for the full story
Rising Above the Clouds: Google’s GenAI Revolution
Google Cloud Next announcements launch salvo of upgrades and partnerships
What is: Putting its impressive torque and talent on display, Google used its flagship cloud conference, Google Cloud Next (GCN), to showcase its latest advancements in AI, infrastructure upgrades, and strategic partnerships
What it means: Google is aggressively positioning itself as a major player in the GenAI market. With 70% of GenAI Unicorns and 50% of start-ups relying on Google Cloud, this year's conference emphasized Google's advancements in AI for building (Vertex AI), AI for users (Duet AI), and the broader AI ecosystem (Google Cloud). Special attention was given to their collaboration with Nvidia, a significant force in the AI hardware domain.
Why it matters: Demonstrating to the world (customers, partners, investors) that Alphabet has depth and breadth of capabilities in GenAI has been the marching orders since OpenAI turned the tech world on its head ~10 months ago. With widespread integration of GenAI tools across its cloud services and clear value-props in terms of efficiency, Google showed remarkable progress in a short period of time. While there were certainly initial missteps, management seems to have righted the ship.
“𝅘𝅥𝅮 It takes Two Bab-AI 𝅘𝅥𝅯”
Google Workspaces users singing a new tune with Duet AI
What it is: Duet AI, Google's co-pilot-like AI collaboration tool, integrates GenAI into their office productivity suite, Workspaces, offering features like email assistance, calendar management, custom visuals, meeting functionalities (transcripts, teleprompter, action items) , code support, and more.
What it means: Google has expanded access to its GenAI-powered suite, enriching everyday tasks, making them more intuitive, efficient, and intelligent. It’s not just about automation but enhancement, tailoring functionalities to user needs and enabling smoother workflows.
Why it matters: The unyielding, accelerating pace of the highly competitive corporate world necessitates productivity tools for individuals and companies to keep pace. While this can be daunting, Google’s introduction of Duet AI not only puts powerful tools in the hands of practitioners, it also takes a bit of the pressure off, offering a toolkit that users can organically adopt into their workflows without disrupting current routines. Duet AI, as implied by the name, is a GenAI partner, not a solo act.
Who Watches The Watchers? The Alignment Dilemma: Censorship vs. Utility
Recent whitepaper sparks debate over AI model value systems
What it is: A whitepaper published on arxiv titled “The Poison of Alignment” analyzes the potential drawbacks of "Alignment" in AI models, whereby Large Language Models (LLMs, GPT-4) are modified after initial training to align with certain value systems.
What it means: The study suggests that such model alignment can reduce the model's utility, raising questions about the extent and authority of imposing value systems on AI models.The implications of alignment are vast, ranging from regional differences in value systems to the potential stifling of AI's capabilities. Decisions on alignment could shape the future use and trust in AI models globally.
Why it matters: Governance of said models by unsophisticated actors in the political and corporate spheres introduces particular challenges; rather than study the underlying technology and leverage it for common good, there exists a bias in these audiences to opt for ignorance and heavy-handed approaches. Proof points such as this are critical in optimizing our use of GenAI.
Don’t I Know You From Somewhere? Incumbents Become Upstarts in AI Chip War
AMD and Intel step up to challenge Nvidia's AI chip dominance
What it is: The AI chip market is heating up. A few months ago, AMD announced its forthcoming AI chips to ship around year-end. Recently posted Intel Habana GAUDI 2 hardware benchmarks rival Nvidia's flagship H100 chips in certain workloads. The ability to rival, and perhaps even surpass (in some cases) the performance of Nvidia's top silicon would make Intel's 2019 acquisition of Haban Labs particularly prudent.
What it means: Nvidia, which has comfortably held the AI chip throne, now finds itself in the crosshairs of formidable contenders. AMD's relentless push, particularly with its open ROCm accelerated computing language becoming a potential drop-in replacement for Nvidia's CUDA, and Intel’s promising performance benchmarks on the back of its 2019 acquisition of Habana Labs, signal a potential shift in the AI chip power dynamics.
Why it matters: For AI developers and businesses, this burgeoning competition is a boon, potentially supporting a phase of rapid technological evolution, more competitive pricing, and a broader array of chip options tailored for diverse AI tasks. Such a competitive environment doesn't just challenge Nvidia's dominance; it spurs innovation across hardware and software layers.
There’s Gold in Them There Texts - High Valuation Underscores Market’s Faith
AI21 valued at $1.4 Bn with $155 Series C raise
What it is: AI21, a GenAI company, showcased with its Series C raise that market demand remains robust for GenAI ventures with tangible business revenue.
What it means: While pre-revenue companies might enjoy more creative valuations, companies like AI21 with proven niches and solid revenue streams can command the substantial valuations we expected in the tech space, based on fundamentals.
Why it matters: The GenAI space remains nascent. Datapoints like this are critical as the market feels-out appropriate pricing of the potential of an emerging industry that has applications across sectors. It is really the capital markets transactions like these that serve as the best metric of strength and success of GenAI. Often mercurial and hyperbolic, capital markets are, over time, ruthlessly efficient. As the space matures, weak franchises will collapse, and winners will emerge, but the scoreboard this will all be measured on will be the NASDAQ, and the tournaments proclaiming the victors and vanquished will be these capital markets press releases trumpeting financings and M&A.
Whose Data Is It Anyway? The Enterprise Data Ownership Debate
Vendors’ updated approach is music to Enterprise customers’ ears
What it is: A recurring theme of GCN has been an emphasis that enterprise data used vendors’ GenAI platforms will not be used for training core AI models by vendors.
What it means: Tech giants are recognizing and responding to feedback and increasing demands for data privacy and ownership from enterprise clients, some of which simply don’t have the right to pass along the data, others which are incredibly sensitive about client data due to regulatory or legal concerns.
Why it matters: Clear policies on data ownership and usage are crucial in building trust with enterprises. Prior to this clarity it was largely impossible for many enterprises to utilize solutions that could not guarantee data security and commit to restrictions on data usage by vendors. This reflects broader trends which should facilitate wider adoption of GenAI. It does draw an interesting contrast with the treatment of the data of non-corporate individuals, which it seems is largely fair game.
That's it for this week! —Feel free to share the AI Geekly with friends, colleagues, family, etc.
About the Author: Brodie Woods
With over 18 years of capital markets experience as a publishing equities analyst, an investment banker, a CTO, and an AI Strategist leading North American banks and boutiques, I bring a unique perspective to the AI Geekly. This viewpoint is informed by participation in two decades of capital market cycles from the front lines; publication of in-depth research for institutional audiences based on proprietary financial models; execution of hundreds of M&A and financing transactions; leadership roles in planning, implementing, and maintaining of the tech stack for a broker dealer; and, most recently, heading the AI strategy for the Capital Markets division of the eighth-largest commercial bank in North America.
TL;DR (Too Long; Didn't Read): Common internet/email abbreviation indicating that if you don't have time to read the longer content, read this short blurb for the gist.
AI (Artificial Intelligence): A subfield of computer science focused on creating intelligent machines capable of performing tasks that require human intelligence, such as natural language understanding, decision-making, and problem-solving.
Capital Markets: Financial markets for buying and selling equity and debt instruments, often facilitating the raising of capital for companies and governments.
Generative AI: A subset of AI that can generate new content, such as text or images, based on patterns learned from large corpora of data. Particularly useful in a range of applications, seemingly able to mimic more complex elements of human cognition.
LLM (Large Language Models): Machine learning models trained on massive datasets to understand and generate human-like text. These models have applications in chatbots, translation services, and content creation.
Alignment: In the context of AI, refers to the process of ensuring that the behavior of AI models aligns with a criteria, often a rubric of values.
CUDA: Compute Unified Device Architecture is a parallel computing platform and application programming interface (API) model created by Nvidia, with dominant market share in the AI space vs. AMD’s ROCm.
Duet AI: a tool by Google that integrates GenAI capabilities into Google Workspaces, enhancing functionalities like email assistance, calendar management, and more.
Enterprise Market: the market that sells technology solutions and services to large organizations rather than individual consumers. Products are often customized and sold in larger quantities.
GenAI: refers to the next generation of artificial intelligence technologies and platforms. It signifies advancements and innovations beyond traditional AI models and methods.
HIPAA, GDPR, CCPA: These are regulatory standards and laws related to data protection and privacy. HIPAA is specific to health information in the U.S., GDPR is a regulation in EU law on data protection and privacy, and CCPA is a statute for residents of California.
ROCm: Radeon Open Compute, is an open-source software foundation for GPU computing on Linux. Developed by AMD, it provides tools, libraries, and frameworks for GPU computing tasks.
Series C Financing: A venture capital funding round typically used by companies to expand market reach, acquire other businesses, or prepare for future rounds of funding or public trading.
Vertex AI: a managed machine learning (ML) platform provided by Google Cloud. It allows developers and data scientists to build, deploy, and scale ML models faster.
Fine-Tuning: The process of adapting a pre-trained machine learning model for a specific task. Fine-tuning involves additional training on a smaller dataset related to the task.
Valuation: The process of determining the current worth of an asset or a company. In the context of startups, valuation is often estimated based on future potential.
AI21 labs: AI21 is a company focused on developing advanced AI models and technologies. They have garnered attention in the AI community for their innovations and contributions.
Alphabet: Alphabet Inc. is an American multinational conglomerate headquartered in Mountain View, California. PArent company of Google.
AMD (Advanced Micro Devices, Inc.): is an American multinational semiconductor company based in Santa Clara, California, that develops computer processors and related technologies.
AWS (Amazon Web Services): a subsidiary of Amazon providing on-demand cloud computing platforms and APIs on a metered pay-as-you-go basis, offering a broad set of global cloud-based products.
Google (Google Cloud): a suite of cloud computing services that runs on the same infrastructure that Google uses for its end-user products. It offers services in computing, storage, and data analytics.
Hugging Face: a company that provides a platform for natural language processing (NLP) technologies, offering tools and models for various NLP tasks.
Intel: American multinational corporation and technology company headquartered in Santa Clara, California. It is the world's largest semiconductor chip manufacturer based on revenue.
Microsoft (MSFT): Microsoft Corporation is an American multinational technology company. It develops, manufactures, licenses, supports, and sells computer software, consumer electronics, personal computers, and related services.
Nvidia: Nvidia Corporation is an American multinational technology company. It designs graphics processing units (GPUs) for the gaming and professional markets, as well as system on a chip units (SoCs) for the mobile computing and automotive market.
OpenAI: a non-profit research company that promotes and develops safe and beneficial artificial intelligence in the service of humanity. Creators of GPT-3, GPT-4 and ChatGPT.
Sundar Pichai: CEO of Google and its parent company Alphabet. One of the companies on the forefront of GenAI.
Jensen Huang: the co-founder and CEO of Nvidia, a major technology company known for its graphics processing units (GPUs) and contributions to the AI industry.
Lisa Su: Dr. Su is the president and CEO of Advanced Micro Devices (AMD). Under her leadership, AMD has seen significant growth and technological advancements.
Thomas Kurian: CEO of Google Cloud, responsible for the direction and activities of Google's cloud services.