• Synthetic
  • Posts
  • Robot dentists, superintelligence, and the Intel melt down

Robot dentists, superintelligence, and the Intel melt down

Plus, Figure's new robot looks like something out of sci-fi

Author

Steve Brown
August 08, 2024

In partnership with

Subscribe to Synthetic

This week’s most interesting and relevant AI news and analysis

This Week in AI

This week’s Synthetic is a little longer than usual. It includes two new elements: an opinion piece on the turmoil at Intel and a new occasional section, “Robot Corner,” focusing on advances in humanoid robotics. We hope you enjoy it.

If you find Synthetic valuable, please consider sharing it with others.

Watch out, dentists! 🦷 AI and robotics are coming for your job. Perceptive, a Boston-based startup, combined advanced robotics, AI, and 3D volumetric imaging to create a restorative dental procedure robot. They claim their robot can place a crown in just 15 minutes, replacing two one-hour visits to the dental office with a human dentist. Perceptive is partly funded by dentist Ed Zuckerberg, who is perhaps best known for having a famous son, Mark. Yes, that Mark.

Groq makes chips focused entirely on accelerating inference for generative AI applications. Their chips are super fast and consume less power than generalized AI accelerators for both training and inference. Meta Chief Scientist Yann LeCun will serve as a technical advisor, and Stu Pann, a vet of Intel and HP, will become COO. Pann jumped ship from Intel, where he led Intel’s foundry business. Groq is now valued at $2.8 billion.

Intel reported a massive net loss in the second quarter, plans to lay off 15,000 employees, and a shamefully low gross margin of 38.7% (in its heyday, Intel consistently commanded gross margins over 60%.) Intel’s stock had its biggest single-day drop in decades; shareholders are suing the company, employee morale is at rock bottom, Nvidia’s market cap is now 30 times that of Intel ($2.52T vs $83B), and the CEO is being criticized for posting bible quotes on X.

Synthetic’s Opinion
A friend of Synthetic, Intel CEO Pat Gelsinger, has been doing his best to right the Intel ship since he took the reins in 2021. Intel is like an oil tanker—turning it takes time. His ambitious catchup plan for “5 nodes in 4 years” is industry speak to describe an acceleration of the shrinking of transistors in Intel’s manufacturing plants. Normally, a shrink (or new node) takes 2 years, so to catch up with industry-leading chip foundry TSMC, Intel is trying to achieve a decade of shrinks in just 4 years. They seem on track, though yields aren’t quite there yet. On the product side of the house, Intel is shanking it, particularly regarding AI, which led to lower-than-expected demand. Since we began with a nautical theme, let’s continue. Intel has a history of missing boats. It demonstrated a lack of strategic patience and exited the cloud hosting business in 2002 before AWS, Azure, or Google Cloud were even a thing. We might all be using Intel Cloud if the company hadn’t given up too soon. In 2007, Intel failed to win the iPhone business (Intel was invited to bid) and then missed the mobile revolution. In 2017 and 2018, Intel passed on an opportunity to buy an early 15-30% stake in OpenAI. Instead, then-CEO Brian Krzanich, who probably did more damage to Intel than anyone, was distracted trying to build a business around drones and sports. Today, Intel’s AI plans are lackluster and late, and it is in catchup again. The Gaudi line of chips (that compete with Nvidia) came through an acquisition, and the latest Meteor Lake (branded Intel Core Ultra) AI PC chip is late, underpowered, and killing Intel’s margins since big parts of it had to be outsourced to TSMC. Shame on Intel when the only CPU available to meet Microsoft’s new Copilot + specification is Qualcomm’s Snapdragon X. The world needs a strong, capable Intel. Nvidia needs real competition, and the U.S. needs leading-edge domestic chip manufacturing capacity. Intel is investing heavily in leading-edge factories, partly supported by funds from the CHIPS act, which some are trying to criticize in light of Intel’s recent struggles and layoff plan. Strategically, the U.S. needs a world-class domestic chip manufacturing capability, and Intel is the only game in town. Like Boeing, the U.S. government cannot let them fail. They are too strategically important. But something’s got to change here. Heads need to roll, and not just the 15,000 announced last week. Intel’s market cap has halved in five years. We must see resignations from the Intel board and firings at the VP and senior VP levels. There has to be accountability, intolerance for incompetence, and fresh thinking in the company's top ranks. Intel has an amazing legacy, and it’s painful and sad to watch the company flail so badly.

Quick Hits

Videos: Humanity’s Final Invention?

The always great, always funny, always interesting Kurzgesagt channel explores Superintelligence. This video is a fun primer on intelligence, evolution, the history of AI, the potential of automated AI research, superintelligence, and existential risk. It’s all delivered with delightful animations, a cheerful British accent you’d expect to hear coming from the Hitchhiker’s Guide to the Galaxy, and an occasional wink. Best of all, their videos, including this one, are easy for anyone to understand. Enjoy!
(18m 30s)

AI Tech and Innovation

A new type of memory, computational random-access memory (CRAM) performs computations like matrix multiplications and scalar addition (key functions involved in AI) directly in its memory cells. The CRAM approach, documented in this peer-reviewed research paper, was developed at the University of Minnesota. Researchers claim that by reducing the need to transfer data between CPU and memory across a data bus, a process that can be far more energy intensive than the computation itself, they use 1000 to 2500 times less energy versus the classic Von Neumann architecture approach.

Artificial Neural Networks (ANNs) are the foundation of all modern AI. They simulate the neurons and synapses of biological brains and arrange artificial neurons in layers. Each neuron takes input signals from neurons in the previous layer, which are weighted by the strength of their synaptic connection. A new architecture, Kolmogorov-Arnold Networks (KANs), applies a more complex function when summing the weights of other neurons, which offers greater flexibility when learning to represent data and increased accuracy while using fewer parameters. For example, in tests, a traditional ANN with 300,000 parameters (nodes and weights) was outperformed by a KAN with just 200 parameters. With simpler networks, it’s easier to understand what’s happening inside them, aiding transparency and interpretability. The downside of KANs? They take longer per parameter to train because they can’t take advantage of GPUs currently optimized for ANN training.

Using a decade of data from Google Maps, Google researchers are building AI to smooth urban traffic flow by intelligently controlling traffic lights. By reducing stop-and-go by up to 30% at 70 test interchanges around the globe, they have reduced emissions at those locations by up to 10%. 🚦

Robot Corner: Introducing Figure 02 🤖

Holy Moly. Figure, arguably the leading humanoid robot maker, just released a video showcasing their latest robot, creatively named “02.” The new model is much more polished than “01” with mostly hidden wires, a longer battery life, more powerful onboard computing, and improved dexterity, physical strength, and vision. Like “01”, “02” learns by observation and practice, and its primary method of communication and instruction is through voice. As previously reported here in Synthetic, Figure robots were on trial at the BMW production line in Spartanburg, South Carolina. This week, BMW announced those tests were successful.

❝

I think there’s an opportunity to ship billions of robots in the coming decade onto the planet”

Brett Adcock, CEO, Figure (Speaking on 60 Minutes)

AI Insights

Perhaps inspired by scenes from Minority Report, Argentina’s Ministry of Security has announced plans to use AI to predict crimes before they are committed. Their new AI unit plans to use AI to prevent, detect, investigate, and prosecute crime. The move comes after the recent victory of right-wing President Javier Milei, who recently met with several tech leaders from OpenAI, Google, Meta, Tesla, and Apple while in Silicon Valley.

Could an AI determine your intelligence level, political leanings, and sexual preferences just by scanning your face? A Stanford University professor thinks so.

The Visual Capitalist charts and ranks companies by their number of generative AI patents. The top 3 might surprise you. IBM, Alphabet, and Microsoft are the only two U.S. companies in the top 10!

Toolkit for the Future

Some AI-powered apps and services to try….

Make a logo in minutes—no design skills required! Auto-create a brand kit with fonts and colors to complement your new logo and create a brand identity. Instantly create branded marketing materials, then use over 300 templates to customize branded business cards, social profiles, email signatures, and more.

Notion helps you focus on the things that actually matter. Organize information, thoughts, to-do lists, calendars, articles, tasks, notes, and team discussions in one well-organized space. Use Notion AI to create content, translate languages, and ask your notes questions.

AI Strategies & tools that will skyrocket your Marketing ROI by 50% 🚀

You don’t realize it yet, but AI has massive potential for you as a marketer.

This free 3-hour Masterclass on AI & ChatGPT (worth $399) will help you become a master of 20+ AI tools & prompting techniques. Join it now for $0

This is for you if you work in any vertical of marketing– writing, designer, campaign managing, influencer marketing, growth marketing, etc.

Ready to shock your team with a 10x boost in revenue & campaign performance? 🚀

You will join 1 Million+ people who have taken this masterclass to learn how to:

  • Create 100+ content pieces for reels,blogs, from one single long form video

  • Put data tracking & reporting for your campaigns on autopilot

  • Do predictive analysis and optimize your marketing campaigns for better results

  • Personalize customer experiences by leveraging the power of AI

You’ll wish you knew about this FREE AI masterclass sooner (Btw, it’s rated at 9.8/10 ⭐)

Invite the HireLogic AI to your next phone, video, or in-person interview. It analyzes the conversation, provides specific insights, and highlights the interviewee’s strengths, skills, and any potential concerns. It creates a full interview transcript so you can focus on the conversation. Free to try.

Melio is a free and easy-to-use payment solution that helps U.S. small businesses pay any expense by bank transfer or credit card - even where cards are not accepted. No setup or monthly fees.

Focus on the email that matters. Deep Clean automatically identifies old, unimportant emails so you can delete them. Snooze emails until a better time and easily track emails that need a response. Automatically filters unimportant emails into specific folders. It works across every device where you check your email. Get to inbox zero with the award-winning SaneBox.

Recording, editing, and refining audio demands more time than most entrepreneurs can spare. Generate voices, high-quality music, and sound effects efficiently and cost-effectively with a single click. Create multilingual voice tracks in up to 29 languages to reach new markets. Try it for free.

Find phone numbers, emails, and company org charts to reach the people you need to. Fill information gaps in your contacts, auto-generate follow-up emails, and identify new target markets. Get 50 free credits and discover why over 500,000 companies use Seamless to grow their business.