UPP News: What's new in the AI field? (Part 12)
Date
July 31st, 2024
Reading Time
5 mins
Tags cloud
What's news
Subscribe to our newsletter and stay up-to-date with the latest AI news updates.
1. Meta releases its biggest ‘open’ AI model yet
July 23rd, 2024
Source: Meta
Meta has launched its largest open-source AI model, Llama 3.1 405B, with 405 billion parameters. While not the biggest ever, it is the largest recent release, trained using advanced techniques and 16,000 Nvidia H100 GPUs. This model excels at text-based tasks like coding and summarization in multiple languages.
Meta is also working on future models that handle images, videos, and speech. The 405B model has a large context window of 128,000 tokens, allowing it to process and summarize long texts. Smaller models, Llama 3.1 8B and 70B, are aimed at general tasks and have similar context capabilities.
Meta is revising its licensing to encourage the use of its models for developing new AI applications, aiming to expand its influence in the AI field.
2. Level AI applies algorithms to contact center pain points
July 23rd, 2024
Source: TechCrunch
Ashish Nagar, formerly with Amazon’s Alexa team, founded Level AI in 2019 to enhance productivity in contact centers using AI. Level AI provides tools that automate tasks, assess agent performance, and offer insights into customer sentiment. It can give agents real-time hints during interactions and assist in coaching.
The platform addresses concerns about data privacy, integration, and job displacement, allowing organizations to manage their own data retention policies. Customers like Affirm, Penske, and Carta use Level AI, which expects to surpass $50 million in annual recurring revenue within two years.
The contact center software market, valued at $61.07 billion in 2024, is projected to grow to $145.20 billion by 2029. Level AI recently raised $39.4 million in Series C funding to expand its reach and workforce.
3. Adobe releases new Firefly AI tools for Illustrator and Photoshop
July 23rd, 2024
Source: Adobe
Adobe’s latest Firefly tools for Photoshop and Illustrator let users create textures and images with simple prompts. Illustrator’s new “Generative Shape Fill” uses text or style references to add details to shapes. Photoshop’s text-to-image generator is now widely available, allowing image creation from descriptions.
Adobe offers free generative credits with Creative Cloud plans, boosting user upgrades. Since March 2023, Firefly has produced over 9 billion images. Other updates include Illustrator’s Dimension and Mockup tools, and Photoshop’s new Selection and Adjustment Brushes.
4. Google makes its Gemini chatbot faster and more widely available
July 25th, 2024
Source: TechCrunch
Google is updating the free version of its AI chatbot, Gemini, to keep up with competitors like Anthropic and OpenAI. The new Gemini 1.5 Flash model will be available in 40 languages across 230 countries, offering better reasoning and image understanding while also being cheaper to run.
Google is expanding Gemini’s context window to 32,000 tokens, allowing it to handle longer text inputs more effectively. File uploads from Google Drive and local devices will soon be available to all users, making advanced data analysis possible.
To reduce AI hallucinations, Google is introducing a feature that adds links to related web content beneath certain responses. The company is also expanding Gemini's availability in Europe and rolling out a global, teen-focused version with additional safeguards and educational resources.
5. Apple used Google's chips to train two AI models, research paper shows
July 30th, 2024
Source: Reuters
Apple has chosen to use Google’s chips instead of Nvidia’s for key components of its AI software infrastructure. This decision is notable because Nvidia dominates the AI processor market. Apple’s research paper revealed that used Google’s tensor processing units (TPUs) to train its AI model, specifically deploying 2,048 TPUv5p chips for iPhone models and 8,192 TPUv4 processors for server models. Unlike Nvidia, which sells standalone GPUs, Google offers TPUs through its cloud platform.
Apple is introducing some AI features to beta users and had disclosed partial reliance on Google hardware earlier, but the full extent was only revealed in a recent paper. Neither Google nor Nvidia commented on this development. Apple’s stock saw a slight decrease following the announcement.
6. OpenAI enters Google-dominated search market with SearchGPT
July 26th, 2024
Source: OpenAI
OpenAI has launched SearchGPT, an AI-powered search engine with real-time internet access, challenging Google's dominance and competing with Microsoft's Bing and Perplexity. This announcement led to a 3% drop in Alphabet's shares. SearchGPT, in its prototype stage, offers summarized search results with source links and allows follow-up questions. It is being tested by a select group of users and publishers.
OpenAI plans to integrate the best features of SearchGPT into ChatGPT. The company is partnering with publishers like News Corp and The Atlantic and has content licensing agreements with major organizations. This collaboration aims to give publishers tools to manage their content in SearchGPT results, increasing pressure on Google to enhance its search capabilities.
Despite its promise, newer AI search providers face challenges, such as Perplexity's legal issues with publishers like Wired and Forbes. Major search engines like Google's and Microsoft's Bing have been incorporating AI since ChatGPT's launch in November 2022. Google recently introduced AI-powered summaries, but its response to SearchGPT's impact remains uncertain.
7. Google's Vertex AI to use Mistral AI's Codestral
July 24th, 2024
Source: Reuters
Google Cloud announced it will use Mistral AI's Codestral model for its Vertex AI service. This partnership underscores Mistral AI's rapid growth, positioning it as a European alternative to Microsoft-backed OpenAI. Vertex AI is Google's generative AI product, and integrating Codestral highlights Google Cloud's commitment to open and flexible AI ecosystems.
Mistral AI, a Paris-based startup founded in 2023 by former Google Deep Mind and Meta AI scientists, has raised over 1 billion euros in funding. In July, BNP Paribas partnered with Mistral AI to use its large language models in various business areas, including customer support, sales, and IT.
8. JPMorgan launches in-house chatbot as AI-based research analyst.
July 26th, 2024
Source: Reuters
JPMorgan Chase has started using a generative AI product, similar to OpenAI's ChatGPT, for its employees, according to the Financial Times. This AI, named LLM Suite, assists employees in writing, idea generation, and summarizing documents. Currently, about 50,000 employees in the asset and wealth management division have access to it.
The rollout began earlier this year, and the AI aims to perform tasks traditionally done by research analysts. This move follows a similar initiative by Morgan Stanley, which partnered with OpenAI last September to offer a GenAI-powered chatbot for financial advisors.
JPMorgan did not comment on the report.
Tags cloud
Newsletter
DISCOVER MORE
![[Recap] UPP Global Technology JSC Establishing Anniversary](/homepage/news-section/new-4.webp)
LET’S TALK...
Content delivered to your inbox
ENTER YOUR EMAIL
YOU WANT TO...
Hanoi, Vietnam
Web3 Tower, No. 15, Alley 4, Duy Tan, Cau Giay, Hanoi, Vietnam