China Launches Massive ChipTech Fund
28 May 2024: Apple bets big on AI for users, China's AI tutors disrupt US education, pre-bunking fights election misinformation, Softbank $9 Bln/Year AI Investment
Good Morning,
Latest in AI,
1. 1-Minute Roundup
2. Headlines
3. AI Investment Deals
4. Other AI News1-Minute Roundup
Headlines:
China's $47.5B chip fund fuels domestic semiconductor industry amid US tensions.
Apple unveils AI strategy with OpenAI chatbot, on-device and cloud capabilities.
Chinese AI homework apps have gained popularity among US students, raising concerns.
"Prebunking" emerges as a proactive tactic against election misinformation powered by AI.
AI Investment Deals:
Hyperleap AI raises $225K pre-seed for its enterprise generative AI platform.
EnCharge AI seeks $70M to develop ultra-efficient AI chips backed by DARPA.
Elon Musk's xAI raises $6B Series B at $24B valuation from top VCs.
AbleCredit raises $1.25M seed to scale AI credit underwriting, offering 20x faster underwriting.
EcoRatings AI cleantech startup raises $1M pre-seed from Google and other investors.
Flam mixed reality AI startup secures $4.5M pre-Series A led by TurboStart.
Others:
SoftBank eyes $9B yearly AI investments, backing Arm and autonomous driving.
Google scrambles to fix bizarre AI search responses involving glue and rocks.
EU task force tackles ChatGPT's privacy compliance challenges under data laws.
Headlines
China Launches $47.5 Billion ChipTech Fund to Boost Domestic Industry.
China established its largest-ever semiconductor investment fund worth $47.5 billion, backed by state-owned entities, to drive development of the domestic chip industry amid escalating US restrictions.
Why it Matters: The move highlights Beijing's renewed push to build an indigenous chip industry as tensions with the US intensify over advanced technology restrictions.
Numbers:
$47.5 billion (344 billion yuan) fund size.
The largest shareholder is China's Ministry of Finance.
Follows previous "Big Funds" of around $100 billion.
Backstory: China has long aimed to boost domestic chip capabilities through programs like "Made in China 2025", providing state capital to local chipmakers like SMIC.
Zoom In: Major Chinese chip stocks like SMIC and Hua Hong Semiconductor rallied on the fund's announcement, aiming to reduce reliance on imported semiconductors.
Reality Check: The fund escalates the global race between superpowers like the US and China for semiconductor supremacy and technological self-reliance. However, Previous state-backed investments have sometimes yielded breakthroughs, and China still needs to catch up in developing cutting-edge chips to replace US circuitry.Apple Lays Bet on User Base for AI Edge
Apple unveils its AI strategy focused on practical tools for its massive user base. It partners with OpenAI for an advanced chatbot while playing catch-up with on-device and cloud AI capabilities.
Why it Matters: Apple recognises the need to showcase AI capabilities to keep consumers and investors interested, even as rivals like Microsoft and Google have stolen the spotlight with major AI announcements.
Backstory: Apple has been working on Project Greymatter, a set of AI tools integrated into apps like Safari, Photos, and Notes, with a mix of on-device and cloud processing.
Between the Lines: Apple's privacy stance led to initial resistance against cloud AI, which is now addressed by keeping data secure and not building user profiles.
Zoom In: Key AI features include transcribing voice memos, photo retouching, Spotlight search improvements, Siri upgrades, custom emojis, notification summaries, and document summarisation.
How it Works: On-device AI runs locally, while more intensive tasks leverage cloud processing on M2 Ultra chips, determined by software managing workload distribution.
Big Picture: While not leapfrogging rivals, Apple aims to quickly become a significant AI player by integrating new AI capabilities across its massive installed base, aided by a partnership with OpenAI for an advanced chatbot.AI Tutors From China Quietly Reshaping How US Students Study
AI homework helper apps, primarily from Chinese companies, are gaining traction, with US students seeking affordable tutoring. They are shaking up the industry as educators grapple with regulating their use.
Why it Matters: The advent of generative AI enables more equitable access to personalised tutoring but risks plagiarism if misused, forcing a reckoning on AI's role in education.
Numbers:
6 million US downloads for the Chinese app QuestionAI.
12 million for ByteDance's Gauth homework app.
5 of the top 20 US education apps use generative AI.
Zoom In: Popular apps like AnswerAI generate step-by-step solutions from student prompts, improving some users' grades and raising accuracy concerns and potential for misuse.
Context: China's tech firms pivoted to overseas markets like the lucrative US after tutoring crackdowns at home, leveraging shared AI foundations to overcome cultural barriers.Reality Check: AI tutoring has limitations, such as hallucinations and a lack of original knowledge. Some students use multiple AIs or cross-check, while others may over-rely on AI assistance.
How it Works: Apps use techniques like retrieval augmentation to boost accuracy by finetuning LLMs on problem sets. Interactivity and learning behaviour insights aim to enhance personalisation.
Prebunking: The New AI Strategy to Combat Election Misinformation
Election officials worldwide adopt "prebunking" to expose people to weakened misinformation paired with explanations, helping build "mental antibodies" against falsehoods ahead of critical elections as traditional debunking efforts fall short.
Why it Matters: With AI and foreign actors threatening election integrity, traditional reactive misinformation tactics like debunking have proven insufficient, prompting this proactive prebunking approach.
Zoom In:
Google is launching pre-bunking ads in Europe to explain disinformation methods.
Local US efforts like celebrity promotions in Arizona's Maricopa County.
Federal agencies are encouraging state/local investment in pre-bunking.
Backstory: Initiatives to counter election misinformation since 2016 have floundered, leading researchers to conclude it's nearly impossible to counter falsehoods once entrenched online.
Reality Check: Legal battles over alleged censorship have chilled collaboration between government and tech firms on misinformation. However, Prebunking could be the best defence in a year when AI-powered threats jeopardise elections globally, and major platforms roll back misinformation labelling. Some experts caution that prebunking's impact needs to be revised, and reliably spotting falsehoods requires intensive skills-building beyond short videos.AI Investment Deals
Hyperleap AI, based in Hyderabad, launched its Generative AI platform after raising $225K in pre-seed funding. The funds will refine its enterprise-ready AI platform, which was developed at T-Hub’s MATH.
EnCharge AI, backed by DARPA, seeks $70M to develop ultra-efficient AI chips. Founded three years ago, the company has raised over $69M in total funding, with DoD grants exceeding $23M.
Elon Musk’s xAI raises $6 billion in Series B funding, boosting its valuation to $24 billion. Investors include Andreessen Horowitz, Sequoia Capital, and Saudi Prince al-Waleed bin Talal.
AbleCredit raises $1.25M seed round led by Merak Ventures to scale AI operations and business growth. The AI credit underwriting startup increases underwriting speed by 20x and reduces costs by 12x.
EcoRatings, an AI cleantech startup, has raised $1M in pre-seed funding from We Founder Circle, 888 VC, Vinners, Indigram Labs Foundation, and Google.
Flam, a mixed reality AI startup, secures $4.5M in pre-Series A funding led by TurboStart, Twin Ventures, and Alphatron Capital.
Others
SoftBank Targets $9bn Annually for AI Investments
SoftBank is committed to investing up to $9 billion annually in artificial intelligence, aiming to support its subsidiary Arm and drive future growth without overstretching finances.
The Japanese tech group led a $1 billion investment in UK self-driving car startup Wayve, indicating its focus on AI while maintaining a solid balance sheet to handle large-scale deals prudently.
Google Scrambles to Remove Weird AI Search Responses
Google manually disables AI Overviews for specific searches after social media users reported bizarre responses, such as instructions to put glue on pizza or eat rocks.
Despite extensive testing, the messy rollout of AI Overviews highlights challenges in achieving high accuracy in AI-generated answers. Google is pledging swift action to rectify the issues and improve its systems.
EU’s ChatGPT Taskforce Tackles Privacy Compliance Challenges
Though crucial legal issues remain unresolved, the EU’s data protection taskforce has shared preliminary conclusions on how current data protection laws apply to OpenAI’s ChatGPT.
Privacy enforcers scrutinise OpenAI’s legal basis for processing data, emphasising the need for valid consent or legitimate interest. Highlight the importance of transparency and user rights in handling personal data the AI chatbot uses.
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