Google AI: Eat Rock, Put Glue On Pizza
24 May 2024: Google AI's wild claims, India's GenAI boom, China's chatbot hurdles, Meta's news ethics, Spotify's AI DJ, Microsoft Pocket Sized AI Model, AI-generated child abuse material charges.
Good Morning & Happy Weekend,
Latest in AI,
1. 1-Minute Roundup
2. Headlines
3. AI Investment Deals
4. Other AI News
1-Minute Roundup
Headlines:
Google AI falsely claims Obama is Muslim and advises eating rocks and putting glue in pizza.
Indian startups leverage AI for fashion, finance, and customer service, raising $831.5 million.
Chinese chatbots navigate censorship and chip shortages, aiming for AI leadership by 2030.
Meta's AI summarises news without attribution, raising ethical concerns.
AI Investment Deals:
Atropos Health raises $33M in Series B funding.
Superlegal raises $5M in seed funding.
Patronus AI raises $17M in Series A funding.
Orca AI raises $23M in funding.
Skribe.ai raises $3.5M in seed funding.
XponentL receives strategic investment.
Others:
Spotify tests a Spanish version of its AI DJ for broader user engagement.
OpenAI debunks claims of cloning Scarlett Johansson's voice for ChatGPT.
First federal charge for creating AI-generated child sexual abuse material.
Microsoft develops compact AI models for efficient use on smartphones and laptops.
Tech giants offer millions for licensing deals to boost AI-generated video models.
FCC fines Steve Kramer for using AI to clone Biden's voice in illegal robocalls.
Arc Search launches a voice-powered query feature for instant answers via phone calls.
Headlines
Google AI Claims: Obama is Muslim, Eat A Rock Daily, Put Glue on Pizza
Google's new AI-powered search overviews are facing intense scrutiny after making bizarre and inaccurate claims, including that former President Barack Obama is Muslim and advising people to eat one rock daily and 1/8 cup glue to pizza sauce.
Why it Matters: The incidents further highlight the risks of deploying unchecked AI systems that can amplify misinformation and hazardous recommendations at the internet scale.
Zoom In: The Obama claim stemmed from an AI misinterpretation of a book chapter title, while the rock-eating advice originated from the satirical website The Onion.
Big Picture: As AI is infused into search to compete with Microsoft and OpenAI, Google faces immense challenges in controlling AI hallucinations and preventing misinformation.
Reality Check: Unfortunately, The search engine couldn’t distinguish between news and satire. High-profile gaffes like Obama’s and rock-eating claims risk undermining trust in Google's quest to enhance search with AI capabilities. While powerful, the incidents show that current AI remains fallible and requires stringent guardrails to prevent the propagation of harmful falsehoods and recommendations.
GenAI Revolutionises Indian Industries
Indian startups are leveraging generative AI models to revolutionise industries from fashion to financial services, offering customised solutions and transforming traditional business operations.
Zoom In:
NeuralGarage uses GenAI for precise lip-syncing in dubbed films.
NeuroPixel.AI automates fashion cataloguing and enhances virtual try-on experiences.
Gupshup’s AI tools improve customer interactions & reduce operational costs for companies like Tata Capital. Gupshup supports over 45K brands.
174 GenAI startups in India. $831.5 million in total funding.
Big Picture: In India, GenAI is set to impact many sectors beyond its current reach, including building, transportation, and farming. Startups like Flyfish.ai provide tailored financial services, while others streamline talent acquisition and customer service processes for companies like Union Bank, IRCTC, and LIC.
The Race to Build Trustworthy AI Chatbots in China
As China aims to lead in AI by 2030, its tech firms face unique challenges in developing advanced chatbots. They must navigate strict censorship rules while overcoming U.S. chip export controls that limit access to cutting-edge hardware.
Why it Matters: AI leadership is a strategic priority for China, but compliance burdens and hardware constraints risk falling behind the open ecosystems of U.S. companies.
Zoom In: From Xi Jinping's Thought doctrines to pre-approved training data sources, Chinese chatbots must align with core socialist values dictated by regulators.
Reality Check: Some analysts believe Chinese chatbots perform reasonably well despite political limitations and restricted training data access. However, the deepening AI chip deficit poses a critical bottleneck for Chinese firms aiming to develop globally competitive large language models.
Meta AI's News Content Controversy
Meta's new AI tool uses news content to generate summaries without proper attribution, reigniting debates about ethical AI usage and its impact on the news industry.
Zoom In:
Meta AI summarises news stories and headlines from various outlets.
In Canada, it uses blocked news content despite banning news links.
The chatbot often reproduces content verbatim from sources like CBS and NBC News.
Meta AI launched in over a dozen countries. Half of Americans get news from social media. Meta-banned news links in Canada to avoid paying
Big Picture: Meta’s AI strategies raise ethical questions about content usage and compensation. This tension mirrors broader industry debates on AI, copyright, and the future of journalism. The AI's capability to bypass news link bans in Canada highlights potential loopholes and the complexity of AI regulation.
AI Investment Deals
Atropos Health raises $33M Series B funding led by Valtruis to scale AI-powered real-world evidence platform and expand pharma partnerships. Atropos Health's Geneva OS and ChatRWD generate publication-grade studies in minutes, leveraging over 200M de-identified patient records.
Superlegal, an NYC-based AI platform for legal contract review for SMBs, raised $5M in seed funding, led by Aleph and Disruptive AI fund and with participation from Alicorn Venture Capital and Tom Glocer.
Patronus AI, a San Francisco-based automated evaluation and security platform provider, raises $17M in Series A funding led by Notable Capital, bringing total funding to $20M.
Orca AI, a London-based autonomous shipping startup, raises $23M in funding led by OCV Partners and MizMaa Ventures, totalling nearly $40M. Orca AI uses AI navigation to enhance safety and efficiency, reducing close encounters by 33% and fuel consumption by up to 5%.
Skribe.ai raised $3.5M in seed funding. It's a virtual AI platform powered by Zoom designed to empower litigating attorneys to capture and analyse testimony more efficiently through innovative software solutions.
XponentL, a data and AI consultancy company, received a strategic investment from Databricks Ventures and Inoca Capital Partners.
Others
Spotify Experiments with Spanish-Speaking AI DJ
Spotify is developing a Spanish-speaking version of its AI DJ, "DJ Livi," expanding the feature's language capabilities beyond the current English-only DJ X.
OpenAI Clears Johansson Voice Claims
Documents and recordings reveal that OpenAI did not copy Scarlett Johansson’s voice for ChatGPT. Instead, a different actress was hired, debunking Johansson's allegations of voice cloning.
U.S. Charges Man with Creating AI-Generated Child Abuse Images
Federal prosecutors have charged Steven Anderegg of Wisconsin for allegedly using AI software, Stable Diffusion, to create over 13,000 explicit images of children, marking the first federal charge for AI-generated child sexual abuse material.
Pocket-Sized AI Models Could Unlock a New Era of Computing
Microsoft researchers have developed compact AI models, like Phi-3-mini, that can run efficiently on smartphones and laptops.
By being more selective about training data, these smaller models show potential for responsive and private AI applications without relying on cloud computing.
Meta and Google Seek AI Deals with Hollywood Studios
Meta and Google offer "tens of millions of dollars" to Hollywood studios for licensing deals to enhance their AI-generated video models.
Netflix and Disney are interested in other collaborations, while Warner Brothers Discovery may be willing to license some content.
$6M Fine for Robocaller Who Used AI to Clone Biden’s Voice
The FCC has proposed a $6M fine for Steve Kramer, who used AI voice-cloning technology to impersonate President Biden in illegal robocalls during a New Hampshire primary election, aiming to suppress voter turnout.
Arc Search Introduces 'Call Arc' for Quick Voice-Powered Queries
Arc Search has launched a new feature, Call Arc. This feature allows users to get instant answers by making a quick phone call within the app, enhancing the existing Voice Search capability.
Like TheAIML?
Please don’t keep it to yourself. Colleagues and friends can sign up here.
For even more,
Our newsletter archive: Click here
Follow us on Linkedin,
Follow us on 𝕏/Twitter
Your feedback is crucial in helping us improve and deliver the content you want to see. Please Email us