Mistral AI releases new model to rival GPT-4 and its own chat assistant

Pivot 5: 5 stories. 5 minutes a day. 5 days a week.

1. Mistral AI releases new model to rival GPT-4 and its own chat assistant

Paris-based AI startup Mistral AI is launching a LLM called Mistral Large, designed to rival top-tier models like GPT-4 and Claude 2. The company is also launching an alternative to ChatGPT called Le Chat, available in beta. Mistral AI's business model is more similar to OpenAI's, offering Mistral Large through a paid API with usage-based pricing. It currently costs $8 per million of input tokens and $24 per million of output tokens to query Mistral Large.

By default, Mistral AI supports context windows of 32k tokens, which is 20% cheaper than GPT-4 Turbo. The company plans to launch a paid version for enterprise clients, allowing them to define moderation mechanisms. Mistral AI is also partnering with Microsoft to provide Mistral models to Azure customers.

Read the full story here 

2. Google aims to relaunch Gemini AI image tool in a few weeks

Google DeepMind CEO Demis Hassabis plans to relaunch its Gemini AI image tool in the next few weeks. The tool, which began offering image generation through its Gemini AI models earlier this month, was paused last week due to inaccuracies in historical depictions.

Google has taken the feature offline while it fixes the issue, and is hoping to have it back online soon. Alphabet's shares were down 3.5% on Monday afternoon, the biggest drag on the benchmark S&P 500 index. Google has been racing to produce AI software to rival OpenAI's ChatGPT since the launch of OpenAI's ChatGPT in November 2022.

Read the full story here

3. Judge blasts law firm for using ChatGPT to estimate legal costs

New York-based law firm Cuddy Law used the OpenAI chatbot ChatGPT to justify a $113,484.62 bill for a recently won trial. The firm argued that the AI tool provided feedback on how much to charge, a sum the losing side was expected to pay.

NYC federal district judge Paul Englemayer deemed the firm's aggressive fee bid "utterly and unusually unpersuasive." The incident is not the first time law practitioners have been caught using ChatGPT, with previous instances including the firing of lawyer Zachariah Crabill and the slap on the wrist of Steven Schwartz.

Read the full story here 

4. Instagram owner Meta forms team to stop AI from tricking voters

Meta, the tech giant behind WhatsApp and Threads, is launching an EU-specific Elections Operations Centre to identify potential threats and implement specific mitigations in real time. The firm, which also owns WhatsApp and Threads, has invested over $20bn into safety and security since 2016, including a global team of around 40,000 people.

Industry expert Deepak Padmanabhan has noted that the company's generative AI strategy could be seen as "lacking teeth" and that it could struggle to deal with AI-generated images. Meta will bring on board three more partners to help deal with the threat.

Read the full story here

5. OpenAI is launching “Feather,” a new AI platform shrouded in secrecy

OpenAI is launching a new AI platform called "Feather," which is shrouded in secrecy. The platform is designed for data labeling and annotation services, specifically tailored for enterprise partners. The platform could be used by contractors to train AI models by labeling sets of images, audio, code, and text.

Experts speculate that Feather could offer tailored features for diverse needs, lightweight AI models, and automated data processing. Despite the lack of official details, Feather may already be in limited use.

Read the full story here 

Advertise with Pivot 5 to reach influential minds & elevate your brand

Get your brand in front of 50,000+ businesses and professionals who rely on Pivot 5 for daily AI updates. Book future ad spots here.