AI Cloning —Emerging Tech Deep Dive
Welcome back to the AI Geekly, by Brodie Woods. Your curated 5-minute-ish read on the latest developments in the rapidly evolving world of AI.
We have a special treat this week:
👆Click the video above👆 for an amazing deep-dive into AI Cloning Technology. We experiment with HeyGen’s AI Avatar solution that brings together an ensemble of AI models to create convincing visual and voice clones —capable of speaking >40 languages.
Check out the video, we guarantee you’ll love it!
TL;DR - Exec Summary
On Romance and AI; AWS and OpenAI Playing Chess; Microsoft’s Top Gun Moment; Just a Teeny-Tiny Brain Implant; Intel’s New Chips — Lightly Salty
AI Ensemble Model Puts Loser Boyfriends on Notice: In a satirical whitepaper Dr. Tiffany Love lays-out an admittedly impressive experiment to replace her “loser Ex” with an ensemble of AI models. It’s funny, but also… with a bit of effort… it kind of could work?
AWS and Microsoft - a Study in Opposites: You would be hard pressed to find two such dramatically different approaches to the recent AI sea change. Microsoft has been furiously building and investing in GenAI with its $10 Bn infusion into OpenAI, a reworking of Azure to better integrate GenAI solutions (its own and OpenAI’s), and finally the integration of Copilot across its entire product base. AWS on the other hand has been patient capital. Deploying funds strategically, establishing its AWS cloud offerings not as a trumpeted competitor to ChatGPT and others, but as a neutral platform. The recent $4 Bn investment in Anthropic is part of this measured approach.
OpenAI Just Can’t Stay Out of the News: Non-stop news around OpenAI with impending availability of DALL-E 3 and multimodal ChatGPT (voice in/out, images in/out). Reports have emerged that OpenAI’s recently release GPT-3.5-turbo-instruct model can play chess at the level of a good human player (not a master —yet). Finally, the company is looking to raise additional capital, with reports of a valuation for the company of $90 Bn -3x the level from a few months ago.
-Read on for the full story
No Need to “Save the Amazon” —AWS is Doing Great!
Series of announcements and investments reflect careful strategy
What is: AWS announced a $4 Bn strategic investment in Anthropic —major competitor to OpenAI’s GPT models—, the integration of conversational LLM technology into its Alexa smart home, and the general availability of its Bedrock GenAI platform.
What it means: A master class in patience, AWS is essentially letting OpenAI, MSFT, and GOOG duke it out in front of the frothing crowds of the market, selectively making its bets once more data is available. Its Anthropic investment ensures access to a suite of best-in-class GenAI models (cheaper and more developed vs. MSFT). Integration of GenAI into Alexa not only reinvigorates a stagnant space (home assistant), but provides a crucial source of human feedback data to sharpen its solutions.
Why it matters: Readers may have noticed relative silence re:AWS over the past few weeks. Amazon doesn’t spend time on big events singing its own praises like certain of its peers, preferring to let results speak for themselves. Slow-and-steady, while less glamorous, is highly effective in an environment of such rapid iteration. The race is not always to the swift…
Chess, Chatbots and Chump Change — Just Another Week at OpenAI
DALL-E 3, multimodal, $90 Bn valuation, and JPMbots
What it is: Sam Altman and team have been busy:
What it means: Solid execution by OpenAI. The Morgan Stanley use case is a classic—a clever, low-risk way to reduce friction in the system. OpenAI seems to be beating competitors on the race to true multimodal, a feather in its cap, and necessary to gain mindshare before AWS and Google release their own multimodal offerings.
Why it matters: Just four months after its last sale, OpenAI is reportedly looking to raise additional capital, valuing the company at up to $90 Bn, 3x its prior valuation. Needle-moving success stories like the Morgan Stanley use case, cutting-edge releases like DALL-E 3 and multimodal interaction all help position OpenAI at the top of the pack when it comes to innovation, earning its choice valuation. We learned GPT-3.5 can play chess this week —looks like Sam’s pretty good at it too.
Microsoft Copilot: Maverick
Copilot coming to Office Suite and Windows 11
Why it matters: Microsoft’s operating systems and productivity solutions are certainly some of the most ubiquitous in the corporate world. These integrations across Office 365, Bing, and Windows 11 will likely be the first interaction many have with AI —these experiences will be critical for setting some of the tone of AI interaction by the masses.
What it means: Microsoft has invested tremendously in its GenAI tools —CEO Satya Nadella is essentially betting his career on the success of MSFT’s AI aspirations -$10 Bn investment coupled with complete reimaginations of each of the company’s core offerings. There is a lot riding on the success of the platform. We will be putting it through its paces and sharing observations with readers of the Geekly.
“The Future of Romance: Novel Techniques for Replacing your Boyfriend with Generative AI”
It’s come to this…
What it is: A satirical whitepaper delves deep into the human psyche as Dr. Tiffany Love applies the latest in AI technology to answer one of life’s greatest questions: can a woman successfully replace her loser ex-boyfriend with an assortment of modern AI tools?
Why it matters: Quite obviously tongue-in-cheek (Dr. Love is a graduate of Cranberry-Lemon University in PA) the paper does a pretty decent job of showcasing real-life examples of novel ways to use AI tools: training an LLM for specific text or speech tasks (asking Dr. L about her day), image generation (Dr. L uses to keep suitors at bay), and more.
What it means: The ideas proposed in the paper are not really that far-fetched, not the AI boyfriend (although there is a thriving market for that apparently), but the concept that chaining together of multiple AI tools can create cohesive solutions that are greater than the sum of their parts.
Who Will Take the Red Blue Pill?
Neuralink looking for brain-implant volunteers
What it is: Elon Musk-owned Neuralink is moving to the human trial phase of its brain-computer interface, surgical implant robot, and brain-signal translation software, following FDA approval four months ago. Neuralink plans to conduct a five-year study with participants with quadriplegia.
Why it matters: These types of technological developments are critical to observe in the context of AI. As the nexus between the human mind and technology become more intertwined, we must consider and be cautious in how we interact with, and trust AI in more fundamental and direct ways. With AI acting as an intermediary between the human mind and the outside world, ensuring alignment of the AI is vital.
What it means: The possibilities unlocked by emerging technologies can be overwhelming. Don’t let the dizzying pace scare you. Advances in AI and healthcare in particular offer some of the greatest opportunities to create transformational, step-change improvements on par with the advent of penicillin.
“Please Clap” Intel’s New Chips Leave Enthusiasts Salty
New Intel silicon brings AI to the chiplet, but OEM-only
What it is: Intel provided details on its upcoming Meteor Lake chips —radically re-designed SoCs employing a chiplet design (branded as “tile”) approach which redistributes some elements of a traditional CPU across several new tiles and employs a new Neural Processing Unit (NPU) AI chip.
Why it matters: There really hasn’t been much reason to mention Intel the past few months. Their hardware doesn’t really come into play when it comes to training AI (that’s all GPUs) and their competitor, AMD, is the preferred CPU-maker for the AI applications that are CPU-intensive (AMD chips have more cores for parallel compute). The introduction of the NPU is interesting, but is really intended purely for local inference and isn’t designed for training or fine-tuning new models.
What it means: The specifications for the Meteor Lake Ultra i9 chips are impressive, but shockingly the chips won’t be made available for end-users to install into desktops themselves —Intel has been clear that the chips will be exclusively for all-in-one PCs. This limits potential demand as enthusiasts who build powerful PCs will be unable to utilize the upgraded hardware.
Before you go… We have one quick question for you:
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.
Ensemble Model: A method in machine learning where multiple models are trained to solve the same problem and combined to get better results. This approach often leads to more robust and accurate predictions compared to using a single model.
AWS (Amazon Web Services): Amazon's on-demand cloud computing platforms and APIs. It offers reliable, scalable, and inexpensive cloud computing services.
GenAI (Generative AI): Refers to a subset of AI focused on generating new data similar to a given set of training data. It's used in creating realistic data for training models when actual data is scarce or unavailable.
AWS Bedrock: a fully managed service under AWS, enabling access to foundation models from Amazon and third-party providers via an API. It facilitates the development of generative AI applications by simplifying the interaction with these models
Foundation Models: Large machine learning models trained on extensive datasets to understand and generate human-like text. They serve as a robust starting point for developers to create more specific AI applications by fine-tuning or building upon these models (vs. a scratch build).
Azure: Microsoft's public cloud computing platform providing a range of cloud services including solutions for AI, analytics, storage and networking. Users can pick and choose from these services to develop and scale new applications, or run existing applications, in the public cloud.
Copilot: Microsoft Copilot is an AI-powered code completion tool that helps programmers write new code and work with existing code faster. Developed in collaboration with OpenAI, it's based on the GPT-3 model and provides suggestions for code completion.
LLM (Language Model): A type of model that learns to predict the next word in a sequence given the words that came before it. Language models are fundamental in various natural language processing tasks and applications, including translation, summarization, and conversation systems.
NPU (Neural Processing Unit): A type of microprocessor designed to accelerate artificial intelligence applications, enhancing the speed and efficiency of AI computations.
Chiplet: A sub-chip microarchitecture where different silicon dies are interconnected on a single package to create a functional chip. This design allows for more modular and scalable chip architectures.
SoC (System on Chip): An integrated circuit that contains all the necessary hardware and computer circuitry required for a computing system. SoCs are used in a variety of technology products, including mobile phones, tablet computers, and other embedded systems.
Microsoft (MSFT): A multinational technology company known for its software products including operating systems, server applications, productivity applications, and games.
OpenAI: An artificial intelligence research laboratory consisting of world-class AI researchers, with the stated goal of ensuring that artificial general intelligence benefits all of humanity.
Anthropic: A company founded by several ex-OpenAI employees, focused on building safe and robust AI systems, ensuring they are aligned with human values for long-term safety.
Morgan Stanley Wealth Management: A division of Morgan Stanley providing a wide range of financial services including investment management, wealth planning, and income management.
DALL-E 3: An advanced version of OpenAI's DALL-E, which generates images from textual descriptions, showcasing a significant advancement in generative AI technology.
ChatGPT: A model by OpenAI designed for conversational purposes, showcasing the advancements in natural language processing and generation.
Neuralink: A neurotechnology company co-founded by Elon Musk, working on implantable brain–machine interfaces to enable direct communication between the human brain and computers.
Intel (INTC): A multinational technology company known for its microprocessors and other semiconductor components used in computers and servers.
AMD (AMD): Advanced Micro Devices, a multinational semiconductor company known for its central processing units (CPUs), graphics processing units (GPUs), and other hardware components in the computing and graphics markets.
Jeff McMillan: Chief Analytics & Data Officer at Morgan Stanley Wealth Management. His expertise in AI and analytics plays a crucial role in advancing digital transformation within the org.
Sam Altman: CEO of OpenAI, Sam Altman, is at the forefront of AI research and development, with the stated goal to ensure that artificial general intelligence (AGI) benefits all of humanity.
Elon Musk: Co-founder of Neuralink, Elon Musk is known for his ventures aimed at advancing technology, with Neuralink focusing on developing implantable brain–machine interfaces to bridge the gap between humans and computers.
Satya Nadella: As the CEO of Microsoft, Satya Nadella has been instrumental in steering Microsoft towards innovative solutions in cloud computing, AI, and other technological fields, enhancing its products and services for a global audience.