Good morning, it’s Tuesday. AI is decoding ancient Roman scrolls, straining U.S. power grids, and sparking new debates about dating. Meanwhile, NVIDIA inks a $700M deal and doubles down on robotics.
Don’t miss the Forward Future 50—our ultimate roundup of 2024’s top AI stories!
Let’s dive in.
☝️ POWERED BY LANGTRACE
Go from shiny demos to reliable AI products that delight your customers with Langtrace. Check out and star our GitHub for the latest updates and join the community of innovators.
20% discount for Langtrace here: https://langtrace.ai/matthewberman
📜 HISTORY DECODED
AI Decodes Ancient Texts: Rewriting History One Scroll at a Time
The Recap: Artificial intelligence is transforming the study of ancient texts by decoding previously unreadable materials, from charred Roman scrolls to eroded cuneiform tablets. Neural networks are unlocking long-lost histories, offering scholars unprecedented access to data and reshaping how we ask—and answer—questions about the past.
Highlights:
AI techniques revealed 16 columns of Greek text in charred scrolls buried by Mount Vesuvius in AD 79, marking a historic breakthrough in papyrology.
Neural networks like the transformer-based Ithaca model restored missing ancient Greek text with 72% accuracy—nearly triple the success rate of human experts working alone.
AI tools are translating vast Korean royal archives and reconstructing rare ancient languages like Linear B, enabling faster analysis of texts too complex for manual decoding.
Researchers employed convolutional and recurrent neural networks to fill in missing cuneiform symbols and reconstruct fragmented Oracle Bone Scripts in China.
AI-driven CT scans and ink-detection algorithms virtually unwrapped scrolls without physical damage, making sealed texts readable for the first time in millennia.
The Vesuvius Challenge aims to decode entire libraries, potentially revealing the largest influx of ancient knowledge in modern history.
Forward Future Takeaways:
AI is accelerating the study of ancient texts, and our understanding of ancient cultures. By combining human expertise with machine precision, AI tools are unlocking vast archives and enabling historians to examine civilizations with new depth. If current trends continue, AI could reveal unknown philosophies, rewrite historical timelines, and even decipher languages long considered lost. → Read the full article here.
⚡ POWER STRUGGLE
The 19th-Century Technology That Threatens A.I.
The Recap: The United States’ dominance in artificial intelligence could be undermined by its outdated energy infrastructure, which struggles to meet the massive power demands of AI systems. While China rapidly expands its electrical capacity and builds cutting-edge energy infrastructure, America faces bottlenecks caused by regulatory hurdles, underinvestment, and fragmented policies.
Highlights:
Training GPT-4 reportedly consumed as much electricity as thousands of U.S. households annually.
U.S. energy investors face delays of up to seven years to connect 2.6 terawatts of planned capacity—three times the country’s current output.
China built 34 ultra-high-voltage power lines and dozens of nuclear plants in the last decade, outpacing the United States, which added only three nuclear reactors.
Microsoft and Google are turning to decommissioned reactors, coal plants, and gas power to keep up with demand, risking climate commitments.
U.S. energy projects face lengthy bureaucratic reviews, high costs, and local opposition.
Rapid permitting processes and state-backed investments in renewables and nuclear give China a competitive edge in scaling AI capabilities.
Proposals include creating an Energy Acceleration Authority to streamline approvals for clean energy infrastructure and address grid bottlenecks.
Forward Future Takeaways:
America’s AI ambitions are at risk unless it modernizes its outdated energy systems to meet soaring demands. Without reforms to streamline regulatory approvals and accelerate clean energy projects, China’s rapid infrastructure growth could close the AI gap. Future success in AI won’t just depend on chips and algorithms—it will hinge on building a 21st-century power grid capable of sustaining the next wave of technological innovation. → Read the full article here.
🏆 2024 HIGHLIGHTS
Forward Future 50: The Top AI Stories of 2024
#14 Musk’s AI Lawsuits: Musk’s legal challenges influence the broader landscape of AI regulation and accountability.
#13 ‘Open’ DeepSeek: V3, the open-source AI model with 671 billion parameters, outperformed rivals in math, coding, and multilingual benchmarks.
#12 Tesla Optimus: Tesla's Optimus robot debuted, showcasing AI-driven tasks like serving drinks and playing games.
#11 OpenAI Advanced Voice Mode: OpenAI expanded ChatGPT’s Voice Mode to the web, enabling real-time, conversational AI with voice interaction.
#10 ElevenLabs Voice Synthesis: ElevenLabs offered high-quality, AI-powered text-to-speech tools in 32 languages, perfect for audiobooks, videos, and more.
#9 Advancements in Test-Time Compute: AI labs prioritized test-time compute, enhancing performance through dynamic resource use during inference.
#8 Sora Video Generation: OpenAI’s Sora Turbo, a faster AI video model, enabled generation of 1080p videos for up to 20 seconds long.
#7 Google Gemini VEO2: Google unveiled Veo 2 for 4K video generation and Imagen 3 for richer, detailed images, enhancing realism and creative precision.
📻️ Tune in Wednesday for 2024’s Top AI Stories—The Final Countdown!
🛰️ NEWS
Looking Forward
💸 Big Tech Spent $125B on AI Data Centers in 2024: Microsoft, Meta, Google, and Amazon poured billions into AI infrastructure. Microsoft led with fixed AI assets, while Google and Amazon prioritized training models over running them.
🏥 Carecode’s AI Agents Target Healthcare Efficiency: Backed by a16z and QED, Brazil’s Carecode aims to cut costs and boost outcomes by automating appointment tasks via AI, focusing on WhatsApp-friendly solutions for accessibility.
⚔️ Israel Used AI to Target Gaza After Hamas Attack: After the Oct. 7 attack, Israel’s military deployed an AI tool, “Habsora,” to generate hundreds of targets as its database of Hamas sites dwindled, sparking AI’s role in warfare debates.
🔤 Will AI Reshape Human Vocabulary: Generative AI may subtly redefine language by prioritizing certain words and phasing out others, raising concerns about losing human-crafted vocabulary and adopting AI-invented terms.
🔬 RESEARCH PAPERS
Open-Source vs. Closed-Source AI: The Battle for LLM Dominance
Open-Source vs. Closed-Source AI.jpg
Large language models (LLMs) are transforming natural language processing, with closed-source systems like GPT-4 leading in performance but criticized for their lack of transparency and accessibility. In contrast, open-source models such as LLaMA and BLOOM emphasize democratization, enabling community-driven innovation and improved representation of underrepresented languages and domains.
While closed-source models leverage massive datasets and computational power to scale effectively, open-source approaches use techniques like LoRA and instruction-tuning to achieve competitive results with fewer resources. Ethical concerns persist, as closed systems lack external oversight, while open models struggle with standardizing bias audits. → Read the full paper here.
📽️ VIDEO
Google Deepmind CEO's Prediction - Digital Biology
Missed it? Matt explored how AI is transforming biology—from solving protein folding to accelerating drug discovery. He dives into AI’s potential to simulate cells, speed up research, and tackle complex challenges with classical and quantum computing. Get the full scoop! 👇
🧰 TOOLBOX
Simplified AI Development, Streamlined Support, and Effortless Meeting Notes
Athina AI | Streamlining AI Development: Athina AI simplifies AI development with tools for testing, monitoring, and collaboration.
Hiver | AI Customer Support: Hiver unifies customer service with AI tools, automations, and analytics for seamless collaboration and scalable support.
Fireflies.ai | Smart Meeting Notes: Fireflies.ai automates meeting transcription, summaries, and insights, boosting collaboration with AI tools and integrations.
🗒️ FEEDBACK
Help Us Get Better
What did you think of today's newsletter? |
Reply to this email if you have specific feedback to share. We’d love to hear from you.
🤠 THE DAILY BYTE
AI Collision or Comedy? Tesla Cam Captures Unexpected Moment.
self-driving fridge hit by self-driving waymo recorded by self-driving tesla — Elliot Arledge (@elliotarledge) 7:26 AM • Dec 30, 2024
@elliotarledgeView post on XCONNECT
Stay in the Know
Follow us on X for quick daily updates and bite-sized content.
Subscribe to our YouTube channel for in-depth technical analysis.
Prefer using an RSS feed? Add Forward Future to your feed here.
Thanks for reading today’s newsletter. See you next time!
The Forward Future Team
🧑🚀 🧑🚀 🧑🚀 🧑🚀
