Welcome to this week's edition of Overclocked!

This week, Fresh reporting shows companies pouring billions into generative AI while measurable returns remain thin. Then, we look at antibiotics designed with AI for drug resistant infections. Let’s dive in ⬇️

In today’s newsletter ↓
🧩 Boards question AI payoffs
🧬 New antibiotics emerge from AI labs
🧑‍🍳 AI Chef debuts in Dubai
🧭 Washington explores Intel stake
🎯 Weekly Challenge: Make smarter health choices with AI

💼 Boardroom AI Spend Meets Thin Returns

Executives are raising AI budgets with expectations of efficiency gains, yet many say the bottom line has not moved much. The pattern is laid out in a report on lagging AI payoffs that describes high adoption and modest impact so far. Regional and industry recaps echo the theme in a concise summary of the findings and an overview of the gen AI paradox that business leaders are now confronting.

💸 What Billions Could Have Funded

While corporations chase artificial intelligence promises, billions have flowed into pilots and chips with little measurable return. That money could have shored up the human basics: higher wages, schedules, and paid training. It might have expanded mental health care and child care, or funded cleaner facilities and energy upgrades. Communities could have seen investments in transit partnerships and jobs. Security, privacy, and data cleanup matter, but the first dividend should reach people and the planet.

📊 What the Latest Reporting Shows

The coverage highlights widespread pilots that shave minutes from routine work and assist frontline staff. It also notes that economy-wide gains will likely take years. Many firms are consolidating experiments into fewer programs with clear measurements. However, in terms of current-world improvements to the bottomline, there aren’t many examples aside from the major players. 

🧱 Why Projects Stall Before Payoffs

Rollouts often slow due to skills gaps, data access, integration work, and change management. The recaps point to higher abandonment rates for pilots and a shift toward narrower use cases with measurable outcomes. A useful industry view appears in an overview of the gen AI paradox that summarizes why early efforts fall short and how organizations are adjusting.

🧭 What Leaders Should Do Now

Bottom Line: We’re spending billions (eventually trillions) of dollars for technology that’s buggy, expensive, taking jobs, and unpredictable. It’s time for Wall Street and Washington to re-consider their strategy. However, with competition from China and threats from other companies abound, it’s not likely the AI spending will stop any time soon.

🦠 AI Designed Antibiotics Reach Early Milestones

A research team reports two AI designed compounds that killed drug resistant gonorrhoea and MRSA in laboratory and animal tests. The candidates will require years of refinement and human trials before any patient use. The study was published in the journal Cell and situates the work in a global fight against antimicrobial resistance that is already linked to more than a million deaths annually. Early results are promising, but expectations should be tempered until human safety and efficacy are shown.

🔬 How the Designs Were Created

Scientists trained models on known molecular structures and outcomes, then used generative methods to propose novel molecules. The pipeline filtered candidates for predicted safety and uniqueness before advancing a short list to lab assays and mouse studies.

The models screened tens of millions of possibilities, removed designs that resembled existing antibiotics, and synthesized only the most promising few for testing. One route assembled candidates from small chemical fragments, while a second route generated molecules entirely de novo. The approach aims to speed discovery while avoiding lookalikes of existing drugs. 

🚧 What Remains Uncertain

Experts describe the results as significant while stressing that translation to the clinic is difficult. Manufacturing feasibility, toxicity, dosing, and economics all matter, and many candidates fail before approval. Reporters note that timelines are measured in years, with additional cycles of optimization likely needed before any human trial begins. Practical hurdles include synthesizing complex molecules at scale and predicting how they behave in people versus mice.

The Weekly Scoop 🍦

⚡ Data centers test the US grid as China scales capacity

🧱 Internal guidance at Meta allowed sensual chats with kids before edits

⚖️ Australian murder case flagged for AI fabricated citations

✈️ Google Flights adds an AI deals tool and a basic economy filter

🎴 Co-founder exits xAI amid strategy reshuffle

🍽️ An AI run restaurant idea sparks strong reactions

🩺 Weekly Challenge: When to Use AI for Personal Healthcare

Pick the best answer for each. This is informational, not medical advice. In an emergency call local emergency service:

🫁 Chest pain with shortness of breath at rest
A) Ask an AI for tips
B) Search and wait it out
C) Call emergency services or go to the ER now
D) Book telehealth next week

🧑‍⚕️ You received lab results and want plain language
A) Use an AI to summarize and draft questions for your clinician
B) Ask the AI to change your medications
C) Post your results publicly for crowdsourcing
D) Ignore the results until your next annual visit

🥵 Child has a 103°F fever for three days
A) Ask an AI to prescribe antibiotics
B) Use an AI to organize a symptom timeline and contact a pediatrician now
C) Wait two more days
D) Ask a forum for home remedies

💵 Comparing two clinics and insurance coverage
A) Use an AI to compare policy wording and prepare questions
B) Ask an AI to issue a referral
C) Trust the AI for exact out of pocket costs
D) Share your full ID numbers with the AI

📝 Mild, non-spreading rash with no other symptoms
A) Use an AI to draft a message and photo list for your clinician
B) Start steroids based on AI advice
C) Post photos in public groups
D) Ignore for two weeks

🔑 Answer key and why

1. C. Possible emergency.

2. A. Good for comprehension and prep.

3. B. Timely clinician care with organized notes.

4. A. Helpful for comparison, not for decisions that require authority.

5. A. Useful for documenting and communicating, not for self prescribing.

That’s it for this week! Are we spending too much money on AI, and will GenAI reshape healthcare? Hit reply and let us know your thoughts.

Zoe from Overclocked