- Front Research
- Posts
- Tech brief: OpenAI Joins Forces with Stack Overflow to Improve Generative AI Models
Tech brief: OpenAI Joins Forces with Stack Overflow to Improve Generative AI Models
+ 5 must-read tech stories.
Tech brief
OpenAI is partnering with Stack Overflow to enhance the performance of its generative AI models for programming tasks. The collaboration aims to improve OpenAI's capacity to answer programming-related queries, while enabling Stack Overflow to leverage OpenAI's expertise for new AI integrations. Stack Overflow had once banned responses from OpenAI's ChatGPT platform due to spam concerns. However, despite initial resistance within the Stack Overflow community toward generative AI, a significant percentage of developers are now embracing these tools.
Researchers at Meta, Ecole des Ponts ParisTech, and Université Paris-Saclay have proposed an improvement on AI large language models (LLMs) by incorporating multi-token prediction, contrary to the traditional single-token prediction. The novel approach can make these models up to three times faster. Despite the option not fitting all model types and language tasks, it offers substantial advantages, especially in performance on generative tasks. The potential benefits include faster business applications like code completion at minimal or no extra cost.
In April, venture capital financing for cryptocurrencies surpassed $1 billion for two consecutive months, the first since late 2022, says RootData. Major investments include $225 million in Monad, a Paradigm and Coinbase Ventures-backed blockchain, and BlackRock's $47 million in tokenizing company Securitize. More than $100 billion has been invested in blockchain sector since 2014, with biggest venture capital funds focused on blockchain infrastructures and decentralized finance protocols.
Apple's upcoming Safari 18 will introduce an Intelligent Search feature and a Web Eraser. The Intelligent Search will summarize webpages in response to search queries, potentially revolutionizing how users access and interact with online content. The Web Eraser, on the other hand, will enable users to permanently remove text, images, or entire sections of webpages. This move by Apple could significantly impact search engines, the search optimization, and publishing industry.
Microsoft is reportedly developing a large AI language model named MAI-1, focusing on AI independence from OpenAI. The model would contain an enormous 500 billion parameters, and Mustafa Suleyman, the former chief of Google Deepmind, guides its conception. Outperforming GPT-4, this move reflects Microsoft's bid to create not only smaller language models but also state-of-the-art, large ones simultaneously.
Apple has initiated the construction of its own AI servers using the M2 Ultra chip, with plans to introduce M4-powered servers by late 2025. Aimed to enhance the performance of its data centers and future AI tools, Apple's new AI strategy underlines the company's ongoing efforts toward vertical supply chain integration.
Reply