DC25003 AI Computer Programming V01 220725

 Coder beats AI rival in a win for humanity

---

Mark Sellman - Technology Correspondent

---

Przemyslaw Debiak said his victory had left him exhausted

A programmer who beat artificial intelligence in a coding competition has declared that “humanity has prevailed (for now!)”.

Przemyslaw Debiak, known as Psyho, said he was “completely exhausted” and “barely alive” after winning the At- Coder World Tour Finals in Tokyo.

Debiak beat an advanced and specially trained AI model from OpenAI, the company behind ChatGPT — for which he used to work. Sam Altman, the chief executive of OpenAI, was gracious in defeat, posting on X that Debiak had done a “good job”. The company added: “Congrats to the champion for holding us off this time.”

Debiak, 41, was competing in a contest where the goal was to write a program that guided robots across a grid using the fewest possible moves. He said that he only had ten hours of sleep in three days, adding on X: “Honestly, the hype feels kind of bizarre. Never expected so many people would be interested in programming contests.”

The Polish programmer’s victory, which comes with a 500,000 yen (£2,500) prize, will be some reassurance to those who fear for the future job prospects of coders.

AI coding has become one of the hottest areas of the technology. Cursor, which provides AI for programmers, has become one of the fastest growing start-ups in history. More than half of Fortune 500 companies use its services.

Leading companies in the field are also seeking to buy AI coding start-ups such as Windsurf, which had a turbulent few days last week. OpenAI had been expected to buy the company, but the deal fell apart, then Google hired Windsurf’s chief executive and other senior staff, and a rival AI start-up bought what was left.

AI coding has also enabled non-technical people to create software through text prompts, leading to a new trend called “vibe coding”. This year’s AI Index Report from Stanford University said that on one benchmark, AI systems could solve just 4.4 per cent of coding problems in 2023, but by last year that had jumped to 71.7 per cent.

Andrej Karpathy, the OpenAI cofounder, said in 2023, after the launch of ChatGPT: “The hottest new programming language is English.”

Debiak told Business Insider that in cases that require straightforward engineering, AI would beat humans. But in longer contests where programmers do everything from scratch, AI had a harder time. “The longer the contest is, the more chances for humans and less chances for AI,” he said.

A recent study from METR, a research group in California, found that while developers believed using AI tools helped them to perform 20 per cent faster, they actually worked 19 per cent slower. One factor was that human developers found AI-generated code unreliable and ended up devoting extra time to reviewing, testing and fixing it.

But the authors warned that AI coding tools were rapidly improving, so their findings could soon be out of date.

Debiak’s victory is the latest contest pitting humans against machines.

Garry Kasparov, the world chess champion, played two matches against IBM’s computer Deep Blue in 1996 and 1997, winning the first but losing the second. In 2016 an AI program developed by DeepMind called AlphaGo beat the Go world champion, Lee Sedol.

DC25002 Using FutureHouse to investigate fibrosis. V01 210725

 AI is playing an increasingly important role in advancing our understanding, diagnosis, and treatment of pulmonary fibrosis, especially idiopathic pulmonary fibrosis (IPF). Here’s how:

🧠 How AI Is Being Used in Pulmonary Fibrosis Research

1. Early Detection and Diagnosis

AI systems are trained to analyze:

• CT scans and X-rays: Detect subtle patterns of scarring before a human radiologist can.

• Pulmonary function test data: Spot early trends indicating fibrosis.

📌 Example: Deep learning algorithms can detect usual interstitial pneumonia (UIP)—a hallmark of IPF—with high accuracy, often aiding in early diagnosis without invasive biopsies.

2. Predicting Disease Progression

AI models use large datasets to forecast:

• How fast the fibrosis will worsen

• Likelihood of complications like acute exacerbations

• Survival timelines and lung function decline

🛠 Tools like machine learning risk calculators help doctors personalize treatment plans.

3. Drug Discovery and Repurposing

AI platforms like FutureHouse (which you mentioned earlier) and others are:

• Analyzing biomedical literature and databases to propose new drug targets

• Simulating molecular interactions to predict whether compounds will reduce lung scarring

• Recommending repurposing of existing medications for fibrosis treatment

📌 Example: FutureHouse’s agents such as Crow, Owl, and Phoenix can sift through thousands of research papers, patents, and clinical trials to identify previously overlooked drug candidates for fibrotic diseases.

4. Identifying Genetic and Biomarker Links

AI models help:

• Discover biomarkers in blood or tissue that correlate with disease stages

• Map gene-expression patterns associated with susceptibility to IPF

🔬 This supports personalized medicine—tailoring treatments to a patient’s genetic and molecular profile.

5. Clinical Trial Optimization

AI helps:

• Match patients to appropriate trials based on detailed phenotyping

• Predict trial outcomes, enabling more efficient study design

⚖️ This saves time, cuts costs, and improves success rates.

🧪 Real-World Initiatives

• FutureHouse AI: Exploring AI scientists that assist in novel hypothesis generation for fibrotic lung disease.

• Mayo Clinic & NVIDIA collaboration: Developing AI tools for automated lung disease detection from imaging.

• UK Biobank & DeepMind: Using large population datasets to model disease risk including for interstitial lung diseases.

💡 Why It Matters

Pulmonary fibrosis is notoriously hard to diagnose early and challenging to treat. AI is shifting the landscape by:

• Speeding up research

• Improving diagnosis accuracy

• Opening new therapeutic pathways

Would you like an example of a specific AI tool or a case study showing how it was used to improve fibrosis outcomes?

DC25001 The AI Scientist V01 210725

 You’re almost certainly referring to FutureHouse—an AI‐powered nonprofit research lab (often styled FutureHouse, one word) focused on building autonomous agents to accelerate scientific discovery.

🔬 What is FutureHouse?

• Founded in November 2023, FutureHouse was inspired by MIT graduate Sam Rodriques (PhD ’19) and computational chemist Andrew White. The organization aims to build an “AI Scientist” capable of automating key steps in research workflows, amplifying human productivity in scientific discovery over a 10‑year mission plan .

• It operates as a philanthropically funded nonprofit, supported notably by former Google CEO Eric Schmidt .

🧠 What does the platform do?

FutureHouse officially launched its AI Scientist platform on May 1, 2025. It’s accessible via both a web interface and API .

At launch, it introduced four specialized agents:

Agent Primary Capability

Crow Summarizes and retrieves scientific literature quickly

Falcon Performs deep literature reviews across extensive open‑access and proprietary datasets

Owl Checks whether a hypothesis or experiment is novel or previously conducted

Phoenix (Experimental) Plans chemistry experiments and proposes novel compounds based on factors like solubility, cost, and novelty

These agents can be chained together in workflows—for instance, from literature review through hypothesis generation to experimental design .

The platform has already demonstrated use cases such as identifying potential therapeutic candidates for dry age‑related macular degeneration (dAMD), and generating hypotheses related to genes linked to polycystic ovary syndrome (PCOS) and Parkinson’s disease .

🧪 Platform strengths & claims

• FutureHouse reports that its agents outperform PhD-level researchers in benchmarks on tasks like literature retrieval and synthesis, due in part to their transparent reasoning mechanisms .

• The company’s architecture integrates AI engineering with experimental science, running a wet lab in San Francisco to iteratively refine tools based on real-world research workflows .

• The long-term vision is to democratize science: enabling every researcher to scale vertically (i.e. conduct 10×–100× more experiments) and horizontally across disciplines .

🚀 Why it matters

• FutureHouse addresses the modern problem of information overload: millions of scientific papers, countless databases, and thousands of tools, creating a bottleneck that no individual researcher can navigate manually .

• By automating key steps in the research lifecycle—from literature review through to experimental planning—FutureHouse aims to accelerate the pace of meaningful discovery, particularly in biology and chemistry .

👤 Summing Up

FutureHouse stands out as a cutting-edge nonprofit AI research platform that blends artificial intelligence, experimental biology, and chemistry to build specialized agents functioning like digital research assistants. It offers publicly‑accessible tools (platform.futurehouse.org), supported by a mission to augment human scientists’ capabilities with transparent, chainable reasoning agents.

Would you like to explore how it compares with other AI science tools—or more details about the individual agents’ capabilities and use cases?