- Starter AI
- Posts
- Intel’s Lunar Lake, 1X improves Eve, OpenAI’s data quest finishes
Intel’s Lunar Lake, 1X improves Eve, OpenAI’s data quest finishes
Enough with the data, let’s move on to training!
Sthefania
June 04, 2024
Hello, Starters!
Chip-making, robotics, data seeking—these are topics we regularly see when discussing AI, but we're not bored of them yet! This is all part of a journey to more advances!
Here’s what you’ll find today:
Intel presents its Lunar Lake chips for Copilot+
Eve, 1X’s humanoid robot gets more autonomous
Altman says OpenAI has enough data for next-gen AI
Showrunner: A streaming service for AI shows
Microsoft introduces Aurora
And more.
Intel has chosen Computex Taipei to showcase the capabilities of their upcoming Lunar Lake chips, which are landing this fall for Copilot+ AI PCs. Offering 48 tera operations per second of AI performance, these chips surpass their former Meteor Lake chips. They will also be available with 16GB or 32GB of RAM.
However, as the chip-making battle tightens, Intel still has much to catch up on if it wants to be on par with rivals like AMD or Qualcomm. Yet, it's expected that Lunar Lake will appeal to those accustomed to Intel technologies, as the chips potentially offer improved gaming capabilities, better AI performance, and enhanced power efficiency.
1X is an innovative robotics firm backed by OpenAI with a big goal: to deploy general-purpose humanoid robots that benefit humans in their everyday lives. It's a mission they're eager to fulfil, as they showcase new updates of Eve, their smartest working android.
Eve is powered by 1X's AI system, which harnesses Embodied Learning to improve its abilities. In recent footage, we can witness the robot chaining simple tasks and performing actions through voice commands.
Sam Altman stated in an interview at the AI for Good Global Summit that OpenAI has enough data to train its next model. This affirmation doesn't surprise us, as we've been covering how the company has landed deal after deal for training data.
He also discussed the importance of high-quality data when it comes to training models. According to Altman, whether it is synthetic or human, the main problem in training arises when the source data is of low quality. Still, the CEO believes that both OpenAI and other companies need to figure out which types of data and methods are best for training AI systems as they become more powerful.
📺Fable Studio, a San Francisco startup, has announced "Showrunner," a streaming platform that allows users to create AI-generated series using their proprietary AI model trained on public data. Dialogue, characters, and shot types can be controlled easily through prompts.
🌫️To overcome the challenges of climate change and create better weather-prediction tools, researchers at Microsoft have developed Aurora, a groundbreaking foundation model able to analyse vast amounts of atmospheric data, making it easier to forecast extreme weather.
⚡️Quick links
What did you think of today's newsletter? |