Google's Strategy to Retain AI Talent,
Plus: Musk's exposed lies about xAI, Meta's Shift to Open Source AGI, OpenAI's Board Search
Presented by
LaPieza is the leading Headhunting company for digital talent in LATAM with a base of +420,000 candidates. The startup, which uses AI in its matching processes, already works with FEMSA, Unilever and Platzi; and has just launched its Executive Search service.
If your company is looking for digital, tech or C-Level talent, validated by AI and executives, schedule a meeting with its COO
Google Retains AI Talent with Special Stock Amid OpenAI’s Poaching
In an effort to retain its top artificial intelligence researchers, Google is leveraging a special stock compensation program. This move comes as Google faces a talent grab from rival OpenAI, which has been offering multimillion-dollar packages. Google’s DeepMind unit, central to the company's AI initiatives, has granted select researchers restricted stock worth up to millions, vesting over one year—a shorter period compared to typical four-year vesting. This strategy aims to counter OpenAI's lucrative offers, yet it’s unclear how DeepMind selects recipients for this program.
Despite staff reductions at Google, the AI research unit's expenditures have soared, with unallocated corporate costs, mostly from DeepMind, rising by nearly 40% to $1.6 billion in the third quarter. Google’s AI team, comprising about 2,000 employees, is heavily invested in developing Gemini, a response to OpenAI’s GPT-4. Recent layoffs of over 1,000 employees at Google could offset the rising costs in AI talent. However, the number of DeepMind employees benefiting from the special stock compensation remains unknown, with many working on the Gemini project.
OpenAI's aggressive recruitment has led to significant departures from Google's Gemini team. Since October, at least two Gemini leaders and several contributors have joined OpenAI, while others moved to different companies. OpenAI, valued at over $85 billion, has been enticing Google staff with potential high-value share offers. Google’s counter-strategy includes offering publicly tradable stock, a benefit not matched by OpenAI. This ongoing talent competition highlights the intensifying rivalry in AI research between the two tech giants.
Zuckerberg Announces Meta's Shift to Open Source AGI, Aligning FAIR and GenAI Teams
Meta, led by CEO Mark Zuckerberg, revealed an ambitious project to develop open source artificial general intelligence (AGI). Zuckerberg announced via Instagram Reel the fusion of two key AI research teams, FAIR and GenAI. This collaboration aims to achieve comprehensive general intelligence, prioritizing open sourcing to maximize accessibility and benefit.
Meta's focus extends beyond AGI development. Zuckerberg highlighted the training of their latest model, Llama 3, and the construction of a robust computing infrastructure. This includes the integration of 350,000 Nvidia H100s by year's end. Further, he linked this development with Meta's metaverse ambitions, specifically mentioning the role of Ray-Ban smart glasses as a medium for AI interaction in daily life.
This announcement trails OpenAI CEO Sam Altman's remarks at the World Economic Forum in Davos regarding AGI's potential risks. It also contrasts with Meta chief scientist Yann LeCun's skepticism about AGI's imminent arrival. Furthermore, the decision to open source Meta's AGI aligns with recent discussions in the AI community, particularly in light of debates sparked by Anthropic's research on the risks of open models. This strategic move by Meta is set to further intensify the ongoing discourse between open and closed source AI methodologies.
OpenAI's Board Search Highlights Nonprofit vs For-Profit Dynamics, Potential Legal Challenges
OpenAI's pursuit to expand its board faces challenges, with no new directors appointed despite ongoing efforts. Recently, Databricks CEO Ali Ghodsi, known for his focus on open-source AI in contrast to OpenAI's closed-source approach, was approached to join OpenAI's nonprofit board, indicating a shift towards diverse perspectives within the organization.
This approach isn't new for OpenAI's board director Adam D’Angelo, who previously engaged in talks with Anthropic’s CEO Dario Amodei about a potential merger during OpenAI's leadership crisis. These interactions suggest an attempt to balance OpenAI’s profit motives with broader industry insights, possibly as a check on CEO Sam Altman's decisions.
OpenAI's dual structure, balancing nonprofit and for-profit elements, has led to growing tension and scrutiny. A recent call by Public Citizen for an investigation into OpenAI's nonprofit status by California's Attorney General Robert Bonta underscores the complexity of its operational model. If found to be overly influenced by its for-profit arm, OpenAI could face dissolution and asset redistribution under California law.
Adding to the complexity is OpenAI's unique profit-sharing arrangement with employees and investors, which raises questions about the organization's financial future, especially in scenarios where profit generation conflicts with its nonprofit objectives. This structure could lead to discontent among backers and employees, particularly in cases where financial returns are jeopardized, as seen during Altman’s temporary ouster.
---
Elon Musk Denies xAI's $500M Funding Reports Refutes Bloomberg's report claiming his AI startup xAI is in $500 million funding talks, aiming for a $15-20 billion valuation. Contrary to reports of securing investor commitments, Musk labeled these claims as 'fake news' on social media. Despite regulatory filings indicating a $1 billion equity offering plan, Musk consistently denies fundraising activities for xAI. Live Mint
Stability AI's Compact Language Model Breakthrough. The new Stable LM 2 1.6B, is a compact language model surpassing older larger models like Stable LM 3B in performance. This model, focusing on multilingual data, follows the recent launch of Stable Code 3B. Despite its smaller size, it outperforms other models under 2 billion parameters, offering new opportunities for developers in the AI ecosystem. Venture Beat
GT Protocol Revolutionizes Web3 with AI. ChainGPT launches GT Protocol, an AI-powered auto-trading platform for cryptocurrency, through its ChainGPT Pad. The protocol, offering access to CeFi, DeFi, and NFTs, integrates advanced AI for efficient trading and portfolio management. GT Protocol's app facilitates diverse crypto investments, set to launch its $GTAI token on January 25. Tech Bullion
Key GPUs from Apple, AMD, Qualcomm Vulnerable to Data Leaks A critical vulnerability, named LeftoverLocals, has been identified in GPUs from Apple, AMD, and Qualcomm, potentially exposing sensitive AI data. Researchers at Trail of Bits discovered that this flaw allows unauthorized access to GPU memory, compromising data security in devices like iPhones and MacBooks. Despite patches from companies like Apple, many devices remain vulnerable, highlighting the challenge in securing increasingly important GPUs in AI applications. Wired
Salesforce CEO Benioff Criticizes AI Companies for IP Issues, accusing AI firms like OpenAI of using copyrighted content without permission. OpenAI, currently negotiating licensing deals with media companies including Time and the New York Times, disputes these claims, with CEO Sam Altman emphasizing a focus on high-quality, limited data. OpenAI has formed partnerships for content use, notably with the Associated Press and German media company Axel Springer SE. Bloomberg