The Freelance Test AI Flunked

Hello Readers👀 

Curious about the biggest AI moves today? You’re in the right place. In today’s edition, we have:

🤖 AI Agents Fall Short On Real Work

🧩 How to Build Reusable Ad Scenes with LTX Elements

⚡ OpenAI Taps AWS for Massive Compute Expansion

🔨Tools and Shifts you cannot miss

🤖 AI Agents Fall Short On Real Work

A new study by Scale AI and the Center for AI Safety reveals that even the most advanced AI agents can only complete a fraction of real freelance projects. The Remote Labor Index tested models on genuine paid assignments to gauge practical automation potential.

The Shift:

1. Real Projects As The Benchmark - The Remote Labor Index used 240 completed Upwork tasks across 23 categories, mirroring real freelance work like design, research, and development. These projects represented over 6,000 hours and $140,000 of verified human labor. 

2. Low Automation Across All Models - Top AI systems, including GPT-5, Sonnet 4.5, and Grok 4, managed just 1–2.5% task completion at professional quality. Most outputs were incomplete, off-brief, or unusable, showing minimal ability to deliver end-to-end results.

3. Limited Performance In Complex Workflows - AI agents failed on projects requiring context, adaptation, and feedback loops, tasks that humans handle naturally. While capable in repetitive or well-defined steps, they falter in dynamic, multi-stage assignments.

The findings provide a measurable baseline for real-world automation readiness. Despite steady progress in reasoning benchmarks, the gap between academic success and applied performance remains wide. The Remote Labor Index highlights how far AI is from replacing skilled human freelancers. 

^{TOGETHER WITH THE AI REPORT}

Practical AI for Business Leaders

The AI Report is the #1 daily read for professionals who want to lead with AI, not get left behind.

You’ll get clear, jargon-free insights you can apply across your business—without needing to be technical.

400,000+ leaders are already subscribed.

👉 Join now and work smarter with AI.

Stay ahead. Stay informed.

🧩 How to Build Reusable Ad Scenes with LTX Elements

LTX’s new Elements feature lets you tag and reuse key visuals, from product shots to actors, across different ad scenes without starting from scratch.

Here’s how to use it:

1. Upload an image of your actor, product, or scene

2. Tag elements (props, faces, logos, backgrounds) once

3. Drop into new shots; they retain the same size, lighting, and style

4. Edit visually: Add, delete, or swap tagged items with precision

5. Animate directly using LTX-2 Fast or Pro models with natural lip sync and motion

It’s like building with brand-safe Lego blocks, fast, flexible, and on-brand every time.

→ Try it here: ltx.studio

⚡ OpenAI Taps AWS for Massive Compute Expansion

OpenAI and Amazon Web Services have entered a seven-year, $38 billion partnership to scale AI workloads globally. The deal expands OpenAI’s compute capacity and signals its first major step beyond Microsoft’s cloud infrastructure.

The Shift:

1. Multi-Year Infrastructure Alliance - The partnership gives OpenAI access to AWS’s powerful UltraServers, built with NVIDIA GB200 and GB300 GPUs. These clusters can scale to tens of millions of CPUs, powering ChatGPT, model training, and agentic workloads with improved efficiency and reliability.

2. Expanding Beyond Microsoft - After renegotiating its deal with Microsoft, OpenAI can now source compute from multiple partners. The AWS agreement is part of a broader $1.4 trillion infrastructure plan that includes collaborations with Oracle, Google, Nvidia, and Broadcom.

3. Scaling for the Next AI Frontier -  All compute will be deployed by late 2026, designed to handle both inference and training at unprecedented scale, reflecting OpenAI’s push to meet growing demand while addressing long-term performance, cost, and sustainability challenges.

The AWS partnership cements OpenAI’s shift toward a multi-cloud future, reducing reliance on a single provider. It strengthens OpenAI’s ability to meet soaring compute demands while preparing for more advanced models. Ultimately, this deal lays the groundwork for scalable, resilient AI infrastructure worldwide.

^{TOGETHER WITH AUTHOR.INC}

Will A Book Grow Your Business?

No one buys a beach house from book sales—they buy it from what the book makes possible.

Author.Inc helps founders turn ideas into world-class books that build revenue, reputation, and reach.

Book a free 15-minute ROI call to see if your book is a go—or a smart wait.

Schedule Your Book ROI Assessment.

🔨AI Tools for the Shift

🧩 Keyword Grid – Analyze any URL and get an instant SEO breakdown of relevance, quality, and intent, the same way Google and ChatGPT interpret it.

🎤 SpeakUp – The world’s first AI-powered platform matching speakers with events, podcasts, and media opportunities, effortlessly and commission-free.

🎨 Agent | Haimeta – Create the future of design using AI-powered tools built to supercharge creativity and digital craftsmanship.

🧠 Rabbitholes AI – Move beyond linear chat apps with an infinite canvas for deep thinking, contextual exploration, and structured creativity.

🎬 Korpi AI – Instantly create viral AI Shorts in trending niches with zero editing skills required.

🚀Quick Shifts

🧠 OpenAI clarifies ChatGPT policy after viral misinformation. The company confirmed there’s no new ban on legal or health advice, saying ChatGPT’s behavior “remains unchanged.” Policies always require professional review for such guidance.

🎨 Studio Ghibli, Bandai Namco, and Square Enix urge OpenAI to halt training on Japanese IP. CODA claims Sora 2’s replication of copyrighted works may breach Japan’s laws and demands prior permission.

🌍 Google Translate is introducing an “Advanced” model on iOS. Users can now choose between “Fast” and “Advanced” modes; the latter handles complex, text-only translations in select languages, marking Google’s deeper AI-first shift.

⚙️ Lambda signs multibillion-dollar AI infrastructure deal with Microsoft. The Nvidia-backed cloud firm will deploy tens of thousands of Nvidia GPUs, including GB300 NVL72 systems, expanding its eight-year partnership as AI compute demand surges.

💾 Microsoft signs $9.7B AI cloud deal with Australia’s IREN.  The five-year agreement secures Nvidia GB300 GPU-powered infrastructure at IREN’s 750MW Texas site, expanding Microsoft’s global AI compute capacity to meet surging demand.

🤳AI Nugget of the Day

That’s all for today’s edition. See you tomorrow as we track down and get you all that matters in the daily AI Shift!

If you loved this edition, let us know how much:

Forward it to your pal to give them a daily dose of the shift so they can 👇