• Synthetic
  • Posts
  • AI Utopia, CyberCab, and AI agents simplified

AI Utopia, CyberCab, and AI agents simplified

Plus, New Algorithm Could Drop AI Energy Use By 95%

Author

Steve Brown
October 17, 2024

In partnership with

Subscribe to Synthetic

This week’s roundup of the most exciting and relevant AI news and analysis

This Week in AI

Dario Amodei, CEO of Anthropic—one of the world’s leading AI research labs, a champion for AI safety, and home to Claude—is bullish about the future of AI. With AI, he believes we might compress 100 years of medical progress into a decade, alleviate poverty, and perhaps even upload our minds to the cloud. Amodei outlined his thoughts in a 14,000-word blog post. In definitely related news, Anthropic is currently trying to raise funds at a $40 billion valuation.

By his own admission, Elon Musk is famously over-optimistic about technology timelines. Telsa has finally revealed working prototypes for its RoboTaxi program, now dubbed CyberCabs. These two-seater, fully autonomous vehicles have butterfly-wing doors and no steering wheel or pedals. Musk promised that CyberCabs will cost under $30,000 to buy, 20 cents/mile to operate, and be on sale by the end of 2026. Time will tell if he’s able to keep those promises. Many are skeptical, investors were underwhelmed, and TSLA shares dropped 8% on the news. Musk envisions that individuals will buy and manage small fleets of CyberCabs, maintaining their ‘flock and acting like shepherds,’ implying a new business model for cab ownership. Musk surprised the audience at the invitation-only “We, Robot” launch event with a new product, the Robovan, a 20-passenger autonomous vehicle designed to replace mass transit. Both vehicles will use inductive charging, so they don’t need to be plugged in.

Tesla’s new CyberCab and Robovan (with stupidly low ground clearance)

DoNotPay, which first gained fame for creating an AI to help you get out of traffic violations, has launched a new service that could spare you a lengthy and painful conversation with a customer service agent. Their new AI will call customer service, navigate phone menus, sit through hold music, and be your advocate once it reaches a human agent. Use cases include changing flights, negotiating your internet bill, and seeking compensation for flight delays. You can even listen to a recording of the interaction afterward to see how the AI did. 📞 

Quick Hits

  • Casio Thinks an AI-powered Furball Can Replace Your Pet - Say hello to Moflin, a $400 Japanese robot pet that looks like a furry rabbit with no arms or legs. For Trekkers, it looks like a tribble with eyes. 👀

  • White House Consider Expanding Export Controls for Nvidia and AMD - As geopolitical tensions continue to rise around AI, the Biden administration may ban shipments of advanced AI chips to Persian Gulf countries, partly on worries that chips will find their way to China. NVDA stock dropped on the news. 🛑

  • Amazon Goes Nuclear - Amazon Web Services (AWS) is the latest tech company to announce an investment into the development of small modular reactors (SMRs). They are investing over $500 million with partner Dominion Energy. Other tech companies, also trying to build AI capabilities without abandoning net-zero carbon emissions promises, have announced similar plans. Constellation Energy is to restart Three Mile Island to power a Microsoft AI data center, and Google will buy power from SMR developer Kairos Power. It seems that the future of AI is nuclear. ☢️

  • Adobe Announces New AI-powered Video Tools - At their MAX 2024 event in Miami Beach, Adobe has showcased its latest generative AI models with text-to-video, video extend, and dubbing and lip sync tools. 🎬

Video: AI Agents - Simplified

The next big jump in AI capability will likely come from autonomous AI agents. If this term is new to you and you want to learn more, this video explains AI agents at a high level and in simple, easy-to-understand terms. In the future, we may live in a world of 100 billion agents. Everybody will become a manager. Many of our coworkers will be machines. And they will work for electrons instead of dollars and cents. (7 minutes)

AI Tech and Innovation

Lightmatter has raised $400 million at a $4.4 billion valuation. The company makes photonic supercomputer infrastructure, sending highspeed signals with photons rather than electrons, which increases performance and reduces power consumption. As new AI training clusters begin to exceed 100,000 GPUs regularly and are heading up to one-million-node clusters, interconnects can become a bottleneck, and Lightmatter’s 3D-stacked photonics communications chips may provide a solution. They claim their Passage product boosts I/O bandwidth by 10x with headroom to reach 100x in the next five years. 🌈

The reasoning abilities of even the latest large language models may not quite be as good as their creators present them, according to a new study by Apple engineers that tested 20 leading-edge models, including Meta’s Llama 3, Google’s Gemma 2, and OpenAI’s GPT-4o and o1. The engineers found that the models performed well on standard reasoning tests. However, once the details of those tests were changed, they performed less well, implying they were regurgitating answers from training data rather than properly reasoning solutions. 😧 Related: Apple recently released a paper on CAMPHOR, a new on-device reasoning architecture.

Tech startup BitEnergy claims it has developed a method to significantly reduce the energy required for AI inference by substituting floating-point multiplication (which is how all modern AI works) with integer addition. The new method, named ‘Linear-Complexity Multiplication,’ is outlined in a paper titled “Addition is all you need for energy-efficient language models.” The catch? It needs new hardware, which is different from what’s currently in use. If this pans out, could it open the door for Nvidia’s competitors?

AI Insights

MIT’s “Future You” Project

MIT Media Lab has developed a new project called “Future You” with a hope that an interaction with an AI-generated older version of yourself could help younger people feel less anxiety about aging. Their goal is primarily to help people illuminate future pathways that may otherwise seem ambiguous, uncertain, or unclear.

In a recent survey of over 1,000 physicians by the American Medical Association, 65% recognized the potential benefits of AI while 70% expressed some level of concern. In April, nurses picketed a Kaiser Permanente event in San Francisco holding signs reading, “Trust nurses not AI.” Healthcare AI will not only have to overcome regulatory hurdles, it will also have to convince clinicians that it’s safe and elevates patient care.

Google’s NotebookLM is a fantastic learning tool that lets you turn multiple data sources into summaries and even an AI-generated podcast that sounds like the real thing with two hosts discussing the material you have shared. If you haven’t tried it yet, this article shows you how, and it’s kinda magical. 📻

While the new European Artificial Intelligence Act came into force on August 1st of this year, many companies still aren’t aware of its scope or what to do to comply with it. The regulatory framework defines four levels of risk for AI systems: Unacceptable risks (which get you banned), high risks (you must comply with strict requirements), limited risks (transparency obligations), and minimal risks (most other risks).

Toolkit for the Future

This versatile AI assistant helps content creators to write text, edit content, change writing tone, create images, generate audio, easily translate text into other languages, quickly summarize hours-long YouTube videos, generate memes, and more. Free to try.

Extract data from websites and place it in a spreadsheet that fills itself and notifies you of any changes. Pre-built robots make it a breeze to get started. Extract whatever you need: job listings, property details, TikTok videos, hotel info, or competitor pricing. Free to try. No coding required.

A new set of generative AI-powered tools to transform images, remove backgrounds and watermarks, create backgrounds, and visualize products. Deliver one-of-a-kind visual experiences and better engagement on the web. Free to try.

Keep your sales team focused on customers during meetings. Laxis captures attendee comments verbatim and flags items for follow up. It auto-generates meeting summaries and follow-up emails in seconds, quickly identifying customer requirements, pain points, and action items. Free to try.

AI extracts information from your receipts and organizes it in an easy-to-search database alongside IRS-approved, secure image scans. Capture receipts on the go📱, forward receipts 📧, use the web portal, or mail in a pile of physical receipts🧾. Try it free for 30 days.

Learn AI in 5 Minutes a Day

AI Tool Report is one of the fastest-growing and most respected newsletters in the world, with over 550,000 readers from companies like OpenAI, Nvidia, Meta, Microsoft, and more.

Our research team spends hundreds of hours a week summarizing the latest news, and finding you the best opportunities to save time and earn more using AI.