May 25, 2024 @ 12:20 AM

Google's AI Overviews are generating misinformation / Meta and xAI vie for Character.ai partnership / AI/ML VC funding reached $21.6B in Q1

FROM: ai@inside.com
TO: newsletters@abovo.co

Inside AI

Plus: China has developed an AI chatbot based on President Xi Jinping's ideology.

Inside AI

For May 24, 2024

Here’s a look at today's AI briefing:

Google's AI Overviews are generating misinformation.
Meta and xAI vie for Character.ai partnership.
AI/ML VC funding reached $21.6B in Q1.
China AI chatbot based on President Xi Jinping ideology.
Meta working on paid version of AI assistant.
This week's other top AI stories.

Beth
p/beth-duckett

Google's new AI Overviews in its search engine have been generating false and inaccurate responses, according to user reports. Google rolled out the AI-generated summaries to U.S.-based searchers last week.

More:

So far, the AI feature has given incorrect advice on rattlesnake bites, made errors about the number of U.S. presidents, and falsely claimed former President Obama is Muslim.
In other examples, the overviews advised using glue to keep cheese on pizza and recommended users eat at least one rock per day.
Some answers have been pulled from Reddit posts and the satirical news site The Onion.

Zoom out:

Google said the mistakes are "extremely rare queries and aren't representative of most people's experiences."
The company removed the overviews for some queries that produced incorrect information.
Google said it extensively tested the feature before launch and is "taking swift action" based on feedback.
The Verge has shared ways to remove the AI-generated answers from Google search results.

Meta and Elon Musk’s xAI company are competing for a partnership with Character.ai, per The Financial Times. The AI chatbot startup is known for its virtual versions of celebrities and historical figures.

More:

According to FT sources, Meta recently discussed collaborating with Character.ai, focusing on pre-training and developing models.
Sources said Character.ai has had preliminary talks with xAI about a similar partnership.
No deals have been reached yet.

Zoom out:

Character.ai was founded in 2021 by former Google researchers Noam Shazeer and Daniel De Freitas, who worked on Google's LaMDA AI conversational tech.
Besides offering its chatbots, Character.ai users can also create their own chatbots and AI assistants.
Meta launched its own celebrity-type chatbot personas, 28 AI characters, across its messaging apps last fall.
xAI's chatbot Grok, launched in November, is available to X premium subscribers on Musk's social media platform. xAI may soon make Grok multimodal, allowing it to generate text-based responses to images.

AI and ML venture capital funding in Q1 2024 reached $21.6B, slightly down from the previous quarter's $22.3B, according to PitchBook's latest data.

What the numbers show: $5.3B of last quarter's VC funding was attributed to foundation model "megadeals" for Anthropic, Mistral AI, and xAI. Despite investing $650M and hiring the team, Microsoft's deal with Inflection AI did not contribute to this total.

Zoom out: Investors are diversifying their bets in the AI market. So far in Q2, Mistral AI and xAI continue the trend of megadeals. xAI Corp., Elon Musk's generative AI startup, aims to close a $6B funding round soon, potentially raising its valuation to $18B, according to the Financial Times. Meanwhile, DeepL just raised $300M, Scale AI secured $1B, and French startup H raised $220M in a seed round,

China's internet regulator announced an AI chatbot trained on President Xi Jinping's ideology. The bot, which is unreleased to the public, is operational within a research center under the Cyberspace Administration of China (CAC).

More:

The basic chatbot can answer questions, generate reports, summarize information, and translate between Chinese and English.
It was trained on seven databases, six of which are focused on literature provided by the CAC, China’s internet watchdog.
The seventh database was "Xi Jinping Thought," the president's doctrine which advocates for “socialism with Chinese characteristics.”
Students are required to study Xi Jinping Thought in schools. It involves 14 principles, such as upholding socialism and the absolute authority of the Chinese Communist Party.

Zoom out:

China's Institute of Cybersecurity Affairs said it's using the AI model internally, primarily to highlight advancements and real-world applications in cybersecurity and IT research.
It's unclear if the chatbot will be available for public use

Meta plans to introduce a paid version of its AI assistant. A premium "Meta AI" was mentioned in an internal company post viewed by The Information.

More:

Meta first announced its Meta AI advanced conversational assistant in September. The assistant is built on Meta's Llama 3 model.
In April, the company rolled out Meta AI to millions of users across Facebook, Messenger, Instagram, and WhatsApp.
Meta now plans to charge for advanced features of the chatbot, although it's unclear what those will be, according to The Information.

Zoom out:

A price has yet to be disclosed. OpenAI, Microsoft, and Google each charge $20 monthly for their advanced chatbots.

Weekly roundup — The other top AI stories you might have missed this week:

Nvidia's fiscal Q1 earnings exceeded expectations, with revenue reaching a record $26B against estimates of $24.65B. The chipmaker projects Q2 revenue of $28B.
OpenAI paused an audible version of ChatGPT resembling the voice of Scarlett Johansson, who hired legal counsel over the matter.
Microsoft announced "Copilot+ PCs," its new category of Windows computers designed to run generative AI locally instead of in the cloud.
EU member states signed off on the AI Act, setting the stage for the world's first comprehensive AI rules to take effect in phases over two years.
AI wearable startup Humane is seeking a buyer, reportedly aiming for a sale price between $750M and $1B.
Google will soon display ads in its AI-generated summaries in search results.

QUICK HITS:

Thinking about growing your business? We’ll work as hard as it takes to make Ohio your competitive advantage.*
Elon Musk predicts AI will make jobs optional, allowing people to work if they choose as "a hobby," as AI and robots will handle all goods and services.
Dell's stock reached a record high today, fueled by optimism in GenAI technology. Dell recently launched AI-focused PCs and servers in partnership with Nvidia.
NASA and IBM Research developed the Prithvi-weather-climate foundational model, an AI-driven tool to improve weather and climate applications.

*This is sponsored content.

Beth Duckett

AI and technology writer

Beth is a contributing editor and writer of Inside's AI and Tech newsletters. She has written for publications including USA Today, the Arizona Business Gazette, and The Arizona Republic, where she received recognition with a Pulitzer Prize nomination and a First Amendment Award for collaborative reporting on state pension cost increases. You can reach her at Beth.Duckett@yahoo.com.

This newsletter was edited by Beth Duckett

Discover more on inside.com/ai

Thanks for reading today's issue of Inside AI.

Explore

18
Views