• Future Perfect
  • Posts
  • #56. πŸ“š Anthropic's Claude AI comprehends entire book in seconds | β˜€οΈ transparent solar cells 1000x more efficient | πŸš— 24/7 AVs in SF by June

#56. πŸ“š Anthropic's Claude AI comprehends entire book in seconds | β˜€οΈ transparent solar cells 1000x more efficient | πŸš— 24/7 AVs in SF by June

Plus: Apple restricts ChatGPT | AI attack map | Modern Hydrogen raises $32.8M | 50%+ of worlds large lakes drying up | Mercedes' EV van platform | Bridgestone's Tesla & Mach-E optimized tire

⚠️ I asked, you voted: Future Perfect is staying as it is and NOT becoming an AI-only newsletter

My last issue of Future Perfect, #55, was an experiment to see whether Future Perfect should go AI-only, i.e., to drop climate change/clean energy and autonomous/electric vehicles news.

Then on Saturday I submitted a poll to validate whether this was in fact what readers wanted, and the results left no room for ambiguity: fully 2/3 of you said Future Perfect should not change to become an AI-only newsletter, and so, at least for the moment, that settles it.

I also lost a few subscribers thanks to that experiment. Wonderful.

Suffice to say, Future Perfect will not be changing, and I'll continue to crank out my same 3:3:3:3 β€” that's 3 topics, 3 articles per topic, 3 cool links, in a 3-minute read summary! β€” every weekday morning at 7AM EST.

Thanks for reading, and please don't forget to spread the word and help me grow my subscriber base!

πŸ€– ARTIFICIAL INTELLIGENCE

Apple has barred employees from using AI tools like OpenAI's ChatGPT and GitHub's AI programming assistant Copilot due to concerns over potential leakage or collection of confidential information. OpenAI typically stores all interactions between users and its AI to improve training and monitor breaches of its terms and services. While a feature to disable chat history has been introduced, interactions are still retained for 30 days, raising fears that sensitive data could potentially be seen by OpenAI moderators or extracted from the system. Not just Apple, but other major companies such as JP Morgan, Verizon, and Amazon have implemented similar restrictions. Interestingly, the ban comes as OpenAI releases a free-to-use iOS app for ChatGPT, supporting voice input and promising an imminent Android version.

As AI technologies rapidly evolve, there is a growing concern about the potential vulnerabilities in AI systems. This report discusses several attack surfaces within an AI system, stressing that the natural language is a primary means of attack. Key components of the AI attack surface include AI Assistants, Agents, Tools, Models, and Storage. The AI Assistants that manage our lives could potentially be attacked, compromising significant personal information and user actions. Agents, or AI-powered entities, could be manipulated to perform actions they weren't supposed to. The Tools used by these Agents may be used for prompt injection attacks. Models have been a traditional focus of AI security, with attacks aiming to make them behave badly, exhibit bias, or subtly skew answers. Storage mechanisms, such as Vector Databases, could be attacked, leading to data leaks or manipulated model results. Looking ahead, it's essential to consider the entire AI-powered ecosystem when planning to attack or defend AI systems, especially where AI intersects with standard business systems.

Anthropic's Claude AI, a competitor to OpenAI's ChatGPT, can reportedly understand a 75,000-word book within seconds, marking a significant advancement in chatbot capabilities. This breakthrough could be particularly beneficial for companies, such as Bloomberg and JP Morgan Chase, that are aiming to utilize AI to gain quicker and more efficient insights in finance and other sectors. The ability of Claude AI to rapidly process large amounts of information is linked to the concept of a 'context window' in computing. In this regard, a token (a fragment of words) is used to streamline data processing, and an LLM's capacity to process tokens at any given time is known as a context window, akin to short-term memory.

πŸ€–πŸ”—πŸ€― COOL AI TOOLS, APPS, VIDEOS, PODCASTS, LINKS, AND MORE!

🌑️ CLIMATE CHANGE & CLEAN ENERGY

Clean energy startup Modern Electron, now renamed Modern Hydrogen, has secured $32.8 million in funding to shift its focus from heat-to-electricity technology to hydrogen fuel production. The Bothell, Washington-based company is developing methane pyrolysis reactors that convert natural gas or biogas into hydrogen, a clean fuel that emits only water vapor upon combustion. Alongside this, the reactors generate a carbon byproduct known as carbon black, which is commonly used in manufacturing goods such as tires and rubber. The company intends to explore potential new uses for carbon black in the construction industry. Uniquely, Modern Hydrogen is focusing on creating smaller, modular devices for decentralized applications to mitigate the cost of transporting hydrogen fuel. The startup was originally established in 2015 and has raised more than $100 million in total. It plans to pilot its technology later this year and hopes to launch commercial products by late 2024 or early 2025.

A groundbreaking study in Science has found that over 50% of the world's largest lakes are losing water, largely due to climate change and unsustainable human consumption. The team, led by Fangfang Yao, a climate fellow at the University of Virginia, utilized 250,000 satellite images from 1992-2020 to analyze 1,972 of Earth's biggest lakes. The results show a staggering 53% global decline in water storage, equivalent to 17 times the volume of Lake Mead, the largest reservoir in the U.S. The study provides insights that could assist water managers and communities in protecting these crucial water sources. They found that sustainable management practices and conservation laws can reverse these trends, as seen with Lake Sevan in Armenia, which has seen an increase in water storage over the last 20 years.

Researchers have developed a highly transparent solar cell using a 2D atomic sheet, achieving an average visible transparency of 79%. These solar cells can potentially be used on building windows, car panels, and even human skin. The transparent solar cells generate a power density of approximately 13 milliwatts per square centimeter, comparable to other transparent solar cell technologies. The potential applications include energy-efficient buildings, power for electric vehicles, and power for personal electronic devices.

πŸš— AUTONOMOUS & ELECTRIC VEHICLES

Mercedes-Benz Vans has announced the development of a new, all-electric VAN.EA platform, designed specifically to meet the diverse needs of its van customers. This modular, scalable architecture will underpin all new Mercedes vans starting from 2026, significantly reducing product complexity and enabling cost efficiencies. The VAN.EA platform is adaptable to various configurations, offering options for short, medium, or long wheelbases, low to high rooflines, and light to heavy-duty applications. Mercedes plans to introduce two variations of the platform, the luxury-oriented VAN.EA-P with a range over 500 km and Level 2 automated driving, and the commercial-focused VAN.EA-C, suited for varied applications from delivery vehicles to ambulances, with Level 4 autonomy in geofenced areas planned for the end of the decade. Mercedes is reorganizing its global production network to accommodate this shift, with all VAN.EA-based vans being electric only and produced in net carbon neutral factories.

Veteran tire manufacturer Bridgestone has unveiled its first electric vehicle (EV)-specific tire, the Turanza EV. This tire is designed to handle the heavier torque of electric vehicles in all seasons, and features Bridgestone's ENLITEN technology, which uses 50% renewable and recycled materials, and a new polymer, PeakLife, for enhanced tread wear resistance. Initially, Bridgestone will target Tesla and Ford EV drivers. The company is planning to release 13 additional tire sizes in 2024.

Cruise and Waymo are likely to receive final approval from the California Public Utilities Commission to operate 24/7 autonomous robotaxi services in San Francisco starting in June. Despite local concerns about safety and transparency, this approval will allow the companies to operate their services without a safety driver present throughout the city. The commission is set to vote on the resolutions on June 29.

πŸŽ‰ THAT’S ALL FOR TODAY!

What'd you think of today's email?

This helps me improve it for you!

Login or Subscribe to participate in polls.

πŸ™ PLEASE:

βœ… Share this post if you liked it

βœ… Share this link (https://www.futureperfect.news) if you're a fan of Future Perfect

βœ… Vote below before leaving!

Thanks for reading!

-Marc πŸ‘‹

Looking for past newsletters? You can find them all here.

Join the conversation

or to participate.