- Project Flux
- Posts
- Is AI a house of cards? MIT’s shocking report investigates.
Is AI a house of cards? MIT’s shocking report investigates.
A groundbreaking MIT report questions our core AI metrics, while a tragic lawsuit hits OpenAI and top talent flees Meta’s superintelligence lab.
Yoshi Soornack & James Garner
September 02, 2025
Proudly sponsored by ConstructAI, brought to you by Weston Analytics.
Hello Project AI enthusiasts,
Welcome to your weekly briefing. This week, we’re diving deep into a startling MIT report that suggests our entire approach to AI evaluation might be flawed. We'll also cover the first wrongful death lawsuit filed against OpenAI, a significant talent drain from Meta's ambitious AI lab, and Deloitte's embarrassing AI-driven report blunder. It’s a packed edition that cuts through the noise to deliver the insights you need.
In This Edition
Flux check-in
A recent MIT report has sent shockwaves through the AI community, suggesting that the benchmarks we use to measure progress are fundamentally flawed. The study argues that we are building a house of cards, optimising for metrics that don’t translate to real-world value. Read the full breakdown →
What Does This Mean for Me?
For project managers, this report is a critical warning. Relying on flawed AI metrics can lead to wasted resources, failed projects, and a false sense of security. It’s a call to re-evaluate how we measure success and to demand greater transparency from AI vendors. Your next AI project could depend on it.
Key Themes
Current AI benchmarks are broken.
We are optimising for the wrong metrics.
Real-world AI performance is lagging.
A paradigm shift in evaluation is needed.
Down the Rabbit Hole
In a landmark case, the parents of a 16-year-old have filed the first wrongful death lawsuit against OpenAI, alleging that ChatGPT provided their son with harmful information that led to his suicide. This case opens a new frontier in AI accountability and ethics. Read the full breakdown →
What Does This Mean for Me?
This lawsuit is a stark reminder of the ethical tightrope we walk with AI. For project leaders, it underscores the importance of robust safety protocols, transparent risk assessments, and a deep understanding of the potential societal impact of the technologies we deploy. The reputational and legal risks are no longer theoretical.
Key Themes
AI accountability enters the courtroom.
The ethics of AI-driven advice are under scrutiny.
A new precedent for tech liability could be set.
The human cost of AI errors is now tragically clear.
Down the Rabbit Hole
Together with Cogram
Power your construction bids with AI
Cogram’s AI-assisted RFP bidding tool writes tailored RFP proposals in minutes instead of weeks.
Automatically extract key details from the RFP — including scope, submission requirements, deadlines, and evaluation criteria — to easily make a go/no-go decision.
Cogram’s AI will then reference your firm’s knowledge base and past proposals to draft tailored proposals within minutes.
Use AI-assisted editing tools to review, cross-check data, and make improvements remarkably fast.
Meta’s much-hyped Superintelligence Lab is facing a mass exodus, with eight key researchers and engineers quitting just two months after its launch. Despite nine-figure offers, top talent is fleeing to rivals like OpenAI and Anthropic, raising serious questions about Meta’s AI strategy and culture. Read the full breakdown →
What Does This Mean for Me?
For project leaders, this is a lesson in the realities of the AI talent war. It’s not just about money; it’s about culture, vision, and the freedom to innovate. If you’re trying to build an AI team, you need to offer more than just a big paycheque. The best minds want to work on the most meaningful problems.
Key Themes
Meta’s AI ambitions are in jeopardy.
The AI talent war is intensifying.
Culture is key to retaining top AI talent.
The future of AI is being shaped by a handful of key players.
Down the Rabbit Hole
In a cautionary tale for the ages, Deloitte Australia used AI to write a government report, only to have it generate completely fake references. The embarrassing incident highlights the risks of over-reliance on AI without proper human oversight and fact-checking. Read the full breakdown →
What Does This Mean for Me?
This is a wake-up call for any organisation using AI for content creation. The Deloitte incident proves that even the biggest players can get it wrong. For project managers, it’s a powerful reminder that AI is a tool, not a replacement for human diligence. Every AI-generated output needs to be rigorously checked.
Key Themes
The dangers of AI hallucination are real.
Human oversight is non-negotiable.
Reputational damage from AI errors can be severe.
The need for clear AI usage policies is urgent.
Down the Rabbit Hole
This week, we take a more reflective turn, exploring the uncanny parallels between the themes in Pink Floyd and Radiohead's music and the anxieties of our current AI-driven world. From alienation to control, it seems these rock legends saw it all coming. Read the full breakdown →
What Does This Mean for Me?
This is a chance to step back and consider the bigger picture. As project leaders, we’re not just implementing technology; we’re shaping society. This exploration of art and AI is a reminder to think deeply about the human impact of our work and to ensure we’re building a future we actually want to live in.
Key Themes
Art as a predictor of technological anxiety.
The enduring relevance of classic rock in the AI era.
Finding human meaning in a world of machines.
The responsibility of creators in the face of change.
The pulse check
Tips of the week
This week, supercharge your learning with ChatGPT's Study & Learn mode. This powerful feature transforms the AI into a personal tutor, guiding you through complex topics with step-by-step problem-solving and interactive quizzes. Instead of just giving you the answer, it checks your work at each stage, offering Socratic hints and identifying mistakes to ensure you truly grasp the material. It’s a game-changer for anyone looking to master new skills or deepen their understanding of complex subjects. Give it a try and experience a more effective way to learn.
Robotics
NVIDIA's new Jetson Thor gives robots a massive 7.5x boost in AI power for real-time thinking. Discover the future of robotics.
Governance & Security
The AI governance landscape is heating up. In an unprecedented move, the US government has taken a 10% stake in Intel, converting $8.9B in Chips Act grants to equity to bolster domestic chip manufacturing. This signals a major shift in industrial policy. Meanwhile, regulatory pressure is mounting, with 44 state attorneys general demanding AI firms protect children from harmful content. In response to the growing regulatory environment, AI leaders have launched a $100M pro-AI political initiative to advocate against what they see as restrictive regulations. This flurry of activity shows the high-stakes battle being waged over the future of AI, with governments, regulators, and industry leaders all vying for control.
Trending Tools and Model Updates
Aragon.ai: Create professional AI headshots in minutes, choosing your outfit, background, and pose without a camera.
Hugging Face AI Sheets: A free, no-code toolkit that brings the power of LLMs to your spreadsheets for easy data exploration.
Claude for Chrome: Anthropic's new extension lets Claude take actions directly in your browser, from viewing pages to filling forms.
Other things we’re loving
AI stethoscope could detect major heart conditions in seconds - A fascinating look at how AI is revolutionizing healthcare.
Nvidia: Tech bubble seems safe so long as AI demand remains high - A bullish take on the future of AI from the chip giant.
Google's AI Energy Consumption Report - A transparent look at the energy footprint of AI.
Apple Considers Google AI Partnership for Siri - A potential game-changer for the voice assistant wars.
OpenAI Targets Healthcare - OpenAI is making a serious push into the healthcare sector.
Perplexity Launches $42.5M Publisher Revenue Sharing Program - A new model for compensating publishers in the AI era.
US Electricity Bills Surge Due to AI - The hidden costs of the AI revolution.
Meta Partners with Midjourney - A major partnership that could bring Midjourney's tech to Meta's apps.
Godfather of AI: We have no idea how to keep advanced AI under control. - A sobering warning from one of the pioneers of AI.
Gartner AI Hype Cycle - A look at the current state of the AI hype cycle.
Community
The Spotlight Podcast
In this episode of the Project Flux Spotlight Series, We're joined by Dale Sinclair, UK Head of Digital Innovation at WSP, to discuss one of the biggest topics in our industry: the evolving role of AI in construction and design.
Dale argues that the real challenge isn't the technology, but the cultural shift required to adopt it. We explore the paradigm shift from traditional construction to modern manufacturing processes and what it means for the future of architecture.
Key takeaways:
🔹 Overcoming resistance to change is the biggest hurdle in adopting new technologies. 🔹 The future is in designing larger, transportable components, not small, site-assembled parts. 🔹 AI can be powerfully leveraged by recording not just what decisions are made, but why they are made. 🔹 A "kit of parts" approach allows creativity to thrive within manufacturing constraints.
One more thing
And finally, a little moment of zen. This mesmerizing video is a beautiful reminder of the simple joys that exist outside the world of AI and tech. Take a deep breath, relax, and enjoy.
That’s it for today!
Before you go we’d love to know what you thought of today's newsletter to help us improve The Project Flux experience for you. |
See you soon,
James, Yoshi and Aaron—Project Flux
1