The Great AI Workforce Reckoning

While Americans celebrate Independence Day, our growing dependence on AI is being intensified in Silicon Valley.

This week, we were greeted with a few developments that highlight the foundations of our AI-driven future: Meta’s $14.3 billion bet on superintelligence, Microsoft's 9,000-person layoff spree, and Europe’s regulatory backbone starting to buckle.

The Billion Dollar Gamble

This week, Meta CEO Mark Zuckerberg sent out an email to his employees, unveiling the company’s new “Superintelligence Labs”, to consolidate the company’s AI ambitions under a single $14.3 billion initiative. And Zuckerberg is ready to spend “hundreds of billions” more on AI.

Read: Mark Zuckerberg Announces Meta ‘Superintelligence’ Effort, More Hires - Bloomberg 

What’s the goal? Same as most Big Tech AI firms - build an artificial general intelligence that matches and surpasses human cognitive abilities across all domains. 

To carry out this ambitious project, Meta is restructuring to bring its foundational AI teams, product divisions, and research units under new leadership and Meta’s recent hire Alexandr Wang (formerly Scale AI's CEO) as Chief AI Officer, and Nat Friedman (ex-GitHub CEO) co-leading applied research. The company has also poached top researchers from rival companies with “signing bonuses as high as $100 million.”

This sounds familiar

The hypothetical AGI or superintelligence is the end goal for pretty much every other AI tech giant. Sam Altman et al. have been running around whispering “AGI” into the ears of squirming investors and venture capitalists for years now. Zuckerberg is now promising “personal superintelligence for everyone.” 🤔❓

But it’s not happening anytime soon. Meta’s own Chief AI scientist Yann LeCun has a differing view. He has iterated many times that we are still far away from reaching a cat or a dog-level intelligence with AI. Human intelligence is still a long shot.

And of course there are risks! Already, the current level of AI is starting to have noticeable impacts on humans. A recent study by MIT Media Lab shows that extensive use of ChatGPT can dull critical thinking, creativity and curiosity. With everyone having superintelligent machines at their disposal, the need to think will go away entirely - what happens to us then, I wonder. Especially considering that LLMs like ChatGPT have been found to deceive and blackmail its own creators just to avoid being replaced.

AI safety experts have also warned that development of superintelligent systems comes with a massive risk—are they going to be aligned with human values, or will they create their own goals? 

Meta’s “superintelligence” announcement comes at a time when the company is simultaneously transitioning to automate 90% of its safety and privacy reviews to AI systems, creating “higher risks” with diminishing human oversight, as per former executives.

So we’re betting big on AI systems—that are ready to deceive their human creators and take over our jobs—to make them more intelligent, while they make us lose ours.

Reminds me of this popular quote that keeps getting misattributed to Albert Einstein, “Artificial intelligence is not a match for natural stupidity.”

Microsoft's Layoff Spree

Microsoft confirmed 9,000 layoffs this week, equal to 4% of its workforce. This announcement comes after 6,000 cuts in May, signifying 15,000 positions being eliminated in the first half of 2025 alone, despite record profitability of $25.8 billion net income for the third quarter of fiscal year 2025.

Months ago, CEO Satya Nadella had already revealed the brutal math behind these layoffs: 20-30% of the company’s code is now AI-generated, with ambitions to reach 95% by 2030. The company is redirecting workforce costs into its $80 billion AI infrastructure investment plans.

Naturally, engineers and software developers are going to be hit hard with such scaling automation by AI tools, but even departments like video gaming (Xbox), sales and customer support are affected in the process of AI-driven simplification.

McKinsey projects 30% of current US jobs could be automated by 2030, while Goldman Sachs estimates up to 50% by 2045.

Baidu's Open-Source Salvo

While Western companies hoard AI capabilities, China's Baidu released its entire Ernie 4.5 model family as open-source. The 10-variant series ranges from 0.3 billion to 424 billion parameters, with the largest model delivering comparable performance to GPT-4.5 at 1% of the cost.

The pricing differential is staggering:

• OpenAI (GPT-4.5): $75 input/$150 output per 1M tokens

• Baidu (Ernie 4.5): $0.55 input/$2.20 output per 1M tokens

Benchmark comparisons show Ernie 4.5's 424B parameter model outperforming DeepSeek V3's 671B system on 22 out of 28 tests, nearly half the size while maintaining superior performance.

Read: Baidu Open Sources ERNIE 4.5: LLM Series Scaling from 0.3B to 424B Parameters - MarkTechPost 

The strategic implications are profound. While companies like Meta and OpenAI are burning billions on closed superintelligence research, Chinese companies are competing by creating democratised versions through open source AI models, potentially accelerating AI adoption across developing markets where OpenAI's pricing remains prohibitive.

Europe's Regulatory Snags

The EU AI Act faces its first major crisis as 45+ European companies—including ASML, Airbus, Mercedes-Benz, and Mistral AI—demand a two-year implementation delay. With critical provisions taking effect on August 2, industry pressure is mounting for regulatory accommodation.

Read: Europe's top CEOs call for Commission to slow down on AI Act | Euronews 

The AI Code of Practice for General-Purpose AI models, originally due May 2, remains unpublished. Without this guidance, companies face legal uncertainty about compliance requirements for technical documentation, copyright alignment, bias testing, and risk assessment.

As political support for delays grows, civil society organisations have strongly criticised such efforts as attempts by Big Tech to weaken EU regulations.

MESSAGE FROM OUR SPONSOR

Learn AI in 5 minutes a day

What’s the secret to staying ahead of the curve in the world of AI? Information. Luckily, you can join 1,000,000+ early adopters reading The Rundown AI — the free newsletter that makes you smarter on AI with just a 5-minute read per day.

Sign up to start learning.

📬 READER FEEDBACK

💬 What are your thoughts on using AI chatbots for therapy? If you have any such experience we would love to hear from you.

Share your thoughts 👉 [email protected]

Have you been a victim of AI?

Have you been scammed by AI-generated videos or audio clips? Did you spot AI-generated nudes of yourself on the internet?

Decode is trying to document cases of abuse of AI, and would like to hear from you. If you are willing to share your experience, do reach out to us at [email protected]. Your privacy is important to us, and we shall preserve your anonymity.

About Decode and Deepfake Watch

Deepfake Watch is an initiative by Decode, dedicated to keeping you abreast of the latest developments in AI and its potential for misuse. Our goal is to foster an informed community capable of challenging digital deceptions and advocating for a transparent digital environment.

We invite you to join the conversation, share your experiences, and contribute to the collective effort to maintain the integrity of our digital landscape. Together, we can build a future where technology amplifies truth, not obscures it.

For inquiries, feedback, or contributions, reach out to us at [email protected].