R ecogni, an AI chip and software startup, introduced a revolutionary computing method on Tuesday that promises to make AI systems smaller, faster, and more cost-effective. This innovation could significantly impact the development and deployment of AI technologies, particularly in data centers and high-performance computing environments.
Recogni, backed by BMW, Bosch, and venture capital firm Mayfield, specializes in creating advanced chips and software tailored for AI inferencing—the process by which trained AI models make predictions or decisions based on new, unseen data. The startup's latest breakthrough, a patented system named Pareto, employs a logarithmic approach that surpasses current methods in running large AI models.
"It is a huge leap in all of the KPIs (key performance indicators) that influence silicon hardware system design when it comes to AI computing," said Gilles Backhus, Recogni's co-founder and VP of AI, in an interview with Reuters.
AI models, like OpenAI's GPT-4 and Google's Gemini, typically require vast amounts of power to perform the complex mathematical operations necessary for tasks such as chatbot interactions. Recogni's Pareto system innovatively converts these multiplication operations into additions, leading to a significant reduction in power consumption while maintaining the accuracy of the AI models.
The effectiveness of Pareto has already been demonstrated on AI models developed by major players like Meta Platforms and Stability AI. Recogni's first chip, developed using Taiwan Semiconductor Manufacturing Co's seven-nanometer process, laid the groundwork for this advancement. The startup is now collaborating with an unnamed partner to make Pareto more widely available, with an official announcement expected in the coming months.
"We are speaking to companies that are putting hardware in data centers and offering it to the world to whoever wants to basically rent it... that's definitely one of the deployment routes that we're considering," Backhus added.
Recogni's novel approach could potentially revolutionize the way AI systems are designed and operated, paving the way for more efficient and accessible AI solutions across various industries.
 |
| Figurines with computers and smartphones are seen in front of the words "Artificial Intelligence AI" in this illustration taken, February 19, 2024. |
0 Comments