2024 July Week 2 AI newsletter

Yiji Suk
4 min readJul 14, 2024

--

1. Salesforce AI Research has released xLAM-1B, a 1B parameter model, which the team claims it to be “the best micro model for function calling, outperforming models 7x its size, including GPT-3.5 & Claude.” (Source: X)

Paper: https://arxiv.org/pdf/2406.18518

Github: https://apigen-pipeline.github.io

Table comparing the respective models’ performance on the Berkeley Function-Calling Benchmark; Overall Accuracy explains the model’s performance on function calling that matches context (higher the better)

2. Skild AI, an AI robotics company building a scalable foundation model for robotics, announced it has closed a $300M Series A funding round. The round was led by Lightspeed Venture Partners, Coatue, SoftBank Group, and Jeff Bezos (through Bezos Expeditions). The funding brings the company to a valuation of $1.5B.

As opposed to vertically designed robots that are built for specific applications, Skild’s model serves as a shared, general-purpose brain for a diverse embodiment of robots, scenarios and tasks, including manipulation, locomotion and navigation. From resilient quadrupeds mastering adverse physical conditions, to vision-based humanoids performing dexterous manipulation of objects for complex household and industrial tasks, the company’s model will enable the use of low-cost robots across a broad range of industries and applications. (Source: BusinessWire)

3. Artificial intelligence robotics startup Standard Bots has raised $63 million over a series of deals. General Catalyst led the round, with participation from the Amazon Industrial Innovation Fund and Samsung Next, a venture arm of Samsung Electronics Co.

Standard Bots builds AI-powered robotic arms that are meant to automate tasks like complex assembly, food preparation or dish washing. Currently, Standard Bots products are used in manufacturing processes, and this new funding will be used to ramp up research and development along with production and hiring.

To train the robots, Standard Bots is developing a transformer-based AI model. The robots observe human demonstrations of the task they are being taught in order to be able to perform it themselves. (Source: Bloomberg)

4. Startup Fireworks AI notched a valuation of $552 million in a funding round led by Sequoia Capital. Sequoia, along with investors including Nvidia Corp., Advanced Micro Devices Inc. and MongoDB Inc. put $52 million into the company in the funding round, brining its total cash haul to $77 million.

Fireworks’ tools aim to make it easier for companies to adopt generative AI, allowing them to access more than 100 models using its platform. The company helps companies fine-tune and customize other models to address specific needs. (Source: Bloomberg)

5. Brain-computer startup Synchron Inc. is tapping OpenAI’s latest artificial intelligence models to help paralyzed patients communicate using its implant.

New York-based Synchron, a competitor to Elon Musk’s brain startup Neuralink Corp., implants its device into the brain via a vein in the neck. Once active, the implant can communicate a patient’s desired actions to a tablet computer. The device, called the Stentrode, can also help patients with tasks like writing emails or answering texts.

The Stentrode is now integrated with GPT-4o, which makes drafting written responses easier for its patients. For example, the technology behind the product can analyze a conversation taking place in the room, gauge the weather and check a calendar in order to suggest a potential response to an email. (Source: Bloomberg)

6. Amazon.com Inc.’s artificially intelligent shopping assistant, Rufus, is now available to all US customers. By tapping an icon in Amazon’s smartphone apps, shoppers can bring up a text chat interface and ask Rufus a range of questions, such as identifying durable outdoor speaker models, comparing running shoes or fetching order updates. (Source: Bloomberg)

7. OpenAI is teaming up with Los Alamos National Laboratory, best known for developing the world’s first atomic bomb, to study the opportunities and risks for using artificial intelligence systems to assist with scientific research.

OpenAI has made a series of announcements in recent weeks — including deals with Moderna Inc. and Color Health — to showcase the potential for its technology to be used for healthcare and biotech. (Source: Bloomberg)

8. OpenAI has come up with a set of five levels to track its progress toward building artificial intelligence software capable of outperforming humans.

The tiers range from the kind of AI available today that can interact in conversational language with people (Level 1) to AI that can do the work of an organization (Level 5).

OpenAI executives told employees that the company believes it is currently on the first level, according to the spokesperson, but on the cusp of reaching the second, which it calls “Reasoners.” This refers to systems that can do basic problem-solving tasks as well as a human with a doctorate-level education who doesn’t have access to any tools. (Source: Bloomberg)

--

--

Yiji Suk
Yiji Suk

Written by Yiji Suk

An enthusiast in modern technology and the future of human civilization. I learn, code, design, and write.