Will AI help us live to 150?

Good morning! We are serving up another piping-hot batch of AI insights—crunchy on the outside, and packed with tasty knowledge on the inside. 🍟

🤯 MYSTERY AI LINK 🤯

(The mystery link can lead to ANYTHING AI-related: tools, memes, articles, videos, and more…)

🔨 AI TOOLS OF THE DAY

🤪 Zeemo: Caption any video in multiple languages. → Check it out

🎨 Avaturn: Turn selfies into AI avatars. → Check it out

👨‍💻 MemeZoo: Create memes from text prompts. → Check it out

ANTHROPIC CEO MAKES BOLD CLAIMS ABOUT AI’S FUTURE 😳

Don’t you find it a bit suspicious that the people making millions of dollars off tech are the ones telling us how great it is? 🙃

What’s up? Anthropic CEO Dario Amodei has released a lengthy, 15,000-word essay titled “Machines of Loving Grace,” which outlines the many ways “AI could transform the world for the better.”

What did he say? Amodei thinks that in the next 7-12 years, AI could help wipe out most infectious diseases, get rid of many cancers, cure genetic disorders, and stop Alzheimer’s early on. He also has a bold take that within 5-10 years, AI could help cure conditions like PTSD, depression, and addiction, either through AI-created drugs or genetic screening (which is pretty controversial). On top of that, he believes AI could help develop drugs that “tune cognitive function and emotional state” to “get [our brains] to behave a bit better and have a more fulfilling day-to-day experience.” Should this happen, Amodei expects the average human lifespan to double to 150. He writes, “My basic prediction is that AI-enabled biology and medicine will allow us to compress the progress that human biologists would have achieved over the next 50-100 years into 5-10 years … I’ll refer to this as the ‘compressed 21st century’: the idea that after powerful AI is developed, we will in a few years make all the progress in biology and medicine that we would have made in the whole 21st century.” In addition to these advancements, Amodei thinks AI has the potential to solve socioeconomic crises like world hunger.

What should you take away? The leaders of big tech companies are incredibly bullish on AI, and they should be—for they are the ones driving innovation. For the average Joe, however, all of these claims can be taken with a grain of salt. AI has made great strides, but don’t quit your job and start shopping for your hover car quite yet.

ADOBE INTRODUCES AI VIDEO CREATION 🎥

I’m about to make so many videos of French fries … 🍟

What’s new? Adobe has introduced AI-powered video generation through its FireFly Video model.

Want the details? Adobe’s latest AI updates for Premiere Pro through Firefly include a video editor called Generative Extend as well as both text-to-video and image-to-video features. The text-to-video feature allows users to generate entire video scenes from textual prompts. For example, you can describe a specific environment or event, and the AI will create the corresponding video content. The image-to-video feature allows users to convert static images into dynamic video scenes. The AI animates elements from the image, enabling a more seamless transition from still visuals to engaging, moving content. Adobe used licensed content through Adobe Stock to train these models, and videos created or edited using Adobe’s Firefly video model can be embedded with Content Credentials to help disclose AI usage and ownership rights when published online.

OPENAI OPEN SOURCES NEW MACHINE LEARNING BENCHMARK 🦾

You must be THIS tall to use AI. 🫳

What’s new? OpenAI researchers introduced MLE-bench, a comprehensive benchmark designed to evaluate AI agents on end-to-end machine learning (ML) engineering tasks.

Why? ML models have demonstrated impressive results in coding tasks, but a significant gap remains in evaluating AI agents’ performance in complex ML engineering. Existing benchmarks typically assess isolated coding skills without considering full-spectrum ML tasks, such as data preparation, model training, and debugging.

How does it work? MLE-bench incorporates 75 curated competitions from Kaggle, spanning diverse fields like natural language processing and computer vision. Human performance on these tasks serves as the baseline for comparison. This benchmark assesses core ML engineering skills, including data preprocessing, model training, and iterative improvement. To no surprise, OpenAI’s new preview-o1 model excelled on the company’s own benchmark tests.

Why is this important? MLE-bench may serve as a valuable tool for assessing AI’s strengths and limitations in ML engineering, providing a more nuanced pathway for further research and innovation. Open sourcing this benchmark encourages collaboration, helping the AI community improve the robustness and reliability of AI agents designed for ML engineering tasks.

TWITTER (X) TUESDAY 🐦

HAS AI REACHED SINGULARITY? CHECK OUT THE FRY METER BELOW: