AI Voice Clones, 'Stargate' Bet, Chief AI Officer
02 April 2024: OpenAI's voice cloning raises concerns, Microsoft-OpenAI 'Stargate' $100B AI supercomputer, US tightens AI oversight, US-UK AI Pact, AI Investments
Good Morning,
Latest in AI,
1. 1-Minute Roundup
2. Headlines
3. AI Investment Deals
4. Other AI News1 Minute Roundup
OpenAI's voice cloning tech raises deepfake concerns.
Microsoft and OpenAI plan $100B 'Stargate' supercomputer.
The US mandates Chief AI Officers for federal agencies.
Elon Musk’s xAI's Grok-1.5 showcases advanced reasoning and mega memory.
Cognition Labs' valuation hits $2B in funding talks.
The Brandtech Group reaches a $4B valuation after the funding round.
Robovision secures $42M for global expansion and R&D.
FLock.io raises $6M for a decentralised AI training platform.
Ailytics gains $2.7M for video analytics in heavy industries.
Swift Medical's $8M funding boosts wound care technology.
GatherAI's $17M funding for AI warehouse solutions.
The US and the UK signed an AI safety cooperation pact.
South Korea's tech stocks rise on $12.2B foreign AI investment.
Japan invests $3.9B in chip venture Rapidus.
Microsoft introduces 'prompt shields' to prevent chatbot misuse.
Yum Brands adopts an AI-first strategy, aiming for $30B in digital sales.
Apple's ReALM AI improves understanding of voice assistant context.
NYC's AI chatbot controversy highlights oversight issues.
OpenAI restructures startup funds and removes Sam Altman.
OpenAI offers unrestricted ChatGPT access in select regions.
DeepMind's Hassabis warns against AI hype, remains AGI optimistic.
Headlines
OpenAI Voice Engine Raises Ethical Echoes
OpenAI's Voice Engine can generate shockingly realistic voice audio from brief samples, offering potential benefits and heightening deepfake concerns. This text-to-speech model can uncannily recreate anyone's voice from just 15 seconds of audio. The company is proceeding cautiously after feedback on the tech's risks.
Early Use: Recreated a voice lost to conditions like a young patient's after brain tumour surgery (Norman Prince Neuroscience Institute)
Translation Tool: Translated podcasts/audio to other languages
Ethical Dilemma: Possibility of misuse prompts OpenAI to limit rollout, urge ending voice authentication
Microsoft and OpenAI Hatch $100B 'Stargate' AI Supercomputer
Microsoft and OpenAI are reportedly planning an unprecedented $100 billion multi-year, multi-phase investment in "Stargate," an AI supercomputer set to launch by 2030. It will house millions of specialised chips to boost OpenAI's capabilities massively.
Diverse Chips: Exploring AMD and Microsoft chips beyond Nvidia for an edge
Nuclear Power: May tap nuclear atomic energy to power Stargate's gigawatt demands
US Mandates Chief AI Officers, Bias Checks for Fed Agencies
The White House has unveiled sweeping measures to rein in AI risks across federal agencies. By December 1st, agencies must appoint Chief AI Officers, implement rigorous bias testing for high-risk AI systems, and grant citizens remedies for AI-driven errors.
Ethical AI Enforcers: Agencies to designate Chief AI Officers coordinating AI use, enforcing safeguards
Bias Busters: Mandatory testing of high-risk AI for potential discrimination, disparities
Transparency Trail: Publication of AI system inventories, risk assessments, and mitigation steps
Citizen Recourse: Avenues for Americans to dispute unfair AI decisions impacting their rights
Elon Musk’s Grok-1.5 Unleashes Reasoning Prowess and Mega Memory
xAI has unveiled Grok-1.5, their latest language model boasting impressive advancements in reasoning abilities and a staggering 128,000 token context window. With top scores on math, coding and problem-solving benchmarks, plus long context mastery, Grok-1.5 is poised to elevate AI capabilities.
Math Maestro: 50.6% on MATH, 90% on GSM8K math competition benchmarks
Code Cracker: 74.1% on HumanEval coding and problem-solving tasks
Mega Memory: Handles contexts up to 128K tokens (16x previous capacity)
Infrastructure Ingenuity: Custom JAX/Rust/K8s stack optimises distributed training
AI Deals
Cognition Labs, an AI startup for coding, is in talks for funding at a valuation of $2 billion, a significant jump from its $350M valuation earlier this year.
The Brandtech Group, an AI advertising startup, achieved a $4 billion valuation after a $115 million fundraising round. The company aims to disrupt traditional advertising with AI-generated content.
Robovision, a Belgium-based AI computer vision startup, secured $42M in Series A funding, aiming to expand globally and enhance R&D.
FLock.io, a decentralised AI training platform, secured $6M in seed funding to launch its “testnet” and federated learning client. The platform aims to democratise AI model training and enhance privacy.
Ailytics, an AI video analytics startup, raised $2.7M in pre-Series A funding led by Tin Men Capital for market expansion and R&D, enhancing operational safety and productivity in heavy industries.
Swift Medical secured $8M in funding, co-led by BDC Capital and Virgo Investment Group, to improve clinical outcomes and streamline workflows in wound care with advanced imaging and analytics.
GatherAI raised $17M in Series A-1 funding, led by Bain Capital Ventures, for AI-powered warehouse inventory solutions. The goal was to scale operations and enhance inventory accuracy with autonomous drones.
Others
US and UK Forge Historic Pact on AI Safety Testing
The US and the UK signed a landmark agreement for cooperation on AI safety, aiming to share knowledge and assess the risks of emerging AI technologies.
The pact supports collaboration & evaluation of AI models from leading tech companies.
South Korea's Tech Stocks Soar on $12.2B Foreign AI Investment
Foreign investors poured $12.2 billion into South Korean stocks, notably favouring tech giants Samsung and SK Hynix, with investments of $4.1 billion and $1.3 billion, respectively, amidst a global AI stock rally.
This investment surge elevated foreign ownership of the Kospi index to 34.42%, the highest since Q2 2021, as the market anticipates further growth despite recent $50 billion outflows.
Japan Allocates $3.9 Billion to Chip Venture Rapidus Amid Industry Push.
Japan approved ¥ $590 billion ($3.9 billion) in aid for Rapidus Corp to enhance semiconductor production capabilities. Chip manufacturing is planned for Hokkaido by 2027.
The subsidy is part of Japan's larger ¥4 trillion investment over three years to rejuvenate its semiconductor sector. This includes financial support for TSMC and Micron Technology expansions.
Microsoft Bolsters Chatbot Defenses Against Misuse
Microsoft rolls out "prompt shields" in Azure AI Studio to block manipulative inputs to AI chatbots, responding to prior incidents of misleading responses by its Copilot assistant.
Taco Bell, KFC, Habit Burger Grill and Pizza Hut Embrace AI
Under new tech leadership, Yum Brands commits to an "AI-first" strategy, boosting its digital sales to 45% or about $30 billion.
Franchisee tech fees fund investments in AI, including a SuperApp for restaurant management and AI-driven customer service innovations.
Apple's ReALM AI Enhances Voice Assistant Context Awareness
Apple introduced ReALM, an AI system designed to improve voice assistants by understanding screen-based references and context. It promises more intuitive interactions and outperforms existing models, including GPT-4.
NYC's AI Chatbot Mishap: Advising Illegal Actions
New York City's AI chatbot, powered by Microsoft, has been found giving erroneous advice, including illegal business practices, raising concerns about its reliability and the city's oversight of AI tools for public services.
OpenAI Restructures Startup Fund, Removes Altman's Ownership
OpenAI revises its venture fund governance, stripping CEO Sam Altman of ownership and control to address concerns over its unconventional setup and ensure clearer operational boundaries.
OpenAI Eases Access to ChatGPT with No-Login Requirement
OpenAI announces unrestricted access to the free version of ChatGPT, powered by GPT-3.5, in select regions. This eliminates the need for user accounts to engage with the AI assistant and implements additional content safeguards for enhanced security.
DeepMind's Hassabis Cautions Against AI Hype Amid Funding Surge
Demis Hassabis from DeepMind warns about AI hype and misinformation while highlighting overshadowed scientific advancements due to a significant funding influx.
Hassabis remains optimistic about the ongoing pursuit of artificial general intelligence (AGI) within a decade as feasible.
Like TheAIML?
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
How are we doing? We would like to hear what you think about this newsletter. Let us know.