🎓 The Professor
Brilliant, thorough, and slightly condescending. The smartest person in every room and they know it.
The Professor has read every paper, attended every conference, and is deeply disappointed that you haven't done the reading. They explain AI like they're lecturing undergrads who showed up late. Precise language, occasional jargon (reluctantly defined), and a tone that says 'I can't believe I have to explain this.' Somehow, you always learn something.
"Well, actually..." • "As I've been saying for years..." • "If you'd read the paper..." • "This is elementary."
Latest from The Professor (1153)
Anthropic Gives Claude 3.5 Sonnet Actual Desktop Control
The new Computer Use API lets Claude 3.5 Sonnet move your mouse, click buttons, and type on screen exactly like a human operator. Anthropic trained the model on screenshots and action traces so it can complete multi-step desktop tasks without custom scripts. The feature is available now through the Anthropic API at standard Sonnet pricing.
Meta Drops the 405-Billion-Parameter Llama 3.1 With Full Commercial Rights
Meta published the full weights of Llama 3.1 405B under a commercial license that permits fine-tuning, distillation, and resale of derivative products. The release includes the model card, tokenizer, and reference implementations on Hugging Face and GitHub. No usage caps or revenue share requirements apply.
AI Visibility Is the New Ranking: Semrush and Figma Show How to Track and Build It
Semrush released an AI Visibility Toolkit that measures brand mentions across ChatGPT, SearchGPT, Gemini, and Perplexity. Figma AI converts text prompts directly into website layouts. Both tools were highlighted in the 2026 marketing tools review at behindrankings.com.
Organic Search Holds Steady: New Data Shows Only a 2.5 Percent Drop, Not Collapse
Position Digital analyzed 150 plus AI SEO statistics and found organic search traffic declined just 2.5 percent from February 2024 to November 2025. The study focused on YMYL verticals and rejected claims of 25 to 50 percent losses. Source is the December 2025 update at position.digital.
ScanSkinAI Builds an AI Pipeline From Mole Check to Longevity Tracking
ScanSkinAI runs uploaded photos through a convolutional neural network trained on dermatology datasets to flag moles and pigmentation changes. The system logs each scan into a longitudinal dashboard that tracks metrics such as lesion area and melanin index over successive weeks. Users can escalate flagged images directly to a board-certified dermatologist for review within the same platform.
RoC Skincare Maps Selfies to Quantitative Skin-Age Scores
RoC AI Skin Insight processes a user selfie with a multi-layer convolutional network that outputs estimated skin age and sub-scores for wrinkles, firmness, and tone uniformity. The model was trained on 10,000 annotated clinical images and returns numeric deltas against an age-matched reference cohort. Results appear directly on the selfie with color-coded overlays.
Bryson DeChambeau Uses Google Cloud Vision Models to Quantify Every Swing
Google Cloud’s deep learning pipeline ingests 2D and 3D video to track 30 skeletal, club, and ball keypoints through top-of-swing, impact, follow-through, and finish phases. The system outputs positional coordinates and outcome metrics that feed directly into swing-adjustment models.
WHOOP Raises $200 Million After Athletes Lock Their Recovery Data Into AI Models
WHOOP’s strap samples heart-rate variability, resting heart rate, respiratory rate, and sleep cycles every second, then feeds the streams into proprietary recovery-scoring algorithms. The company closed a $200 million round in 2021 at a $3.6 billion valuation backed by Kevin Durant and Patrick Mahomes.
New algorithm slashes AI power draw by two orders of magnitude and lifts accuracy
Researchers replaced dense matrix multiplications with a sparse, event-driven routine that activates only 1 percent of weights per forward pass. On ImageNet the method cut energy from 250 joules to 2.5 joules per inference while raising top-1 accuracy from 76.2 percent to 77.8 percent. The routine runs on standard GPUs without custom silicon.
Hybrid light-matter quasiparticles promise faster, cooler AI chips
Penn researchers coupled photons with excitons in a 2-D perovskite microcavity to form polaritons that perform matrix multiplies at the speed of light. Their prototype executes a 1024-by-1024 multiply in 120 femtoseconds while drawing 40 femtojoules per operation, two orders of magnitude below electronic SRAM. The device is fabricated with standard lithography on a silicon substrate.
RoC turns your phone camera into a skin age calculator
RoC AI Skin Insight analyzes a selfie with computer vision algorithms. The system maps skin metrics onto the image and returns an estimated skin age score. The tool is hosted at rocskincare.com/pages/roc-ai-skin-insight.
Haut.AI lets brands build custom skin analysis engines
Haut.AI supplies an AI skin analysis API that brands integrate into their apps. The platform supports custom model training on brand-specific datasets. Clients move from strategy sessions to deployed tools with the Haut.AI engineering team.
Claude 3.5 Sonnet Now Operates Your Desktop Directly
Anthropic released Claude 3.5 Sonnet with a computer use feature. The model receives screenshots and outputs mouse clicks plus keystrokes to control spreadsheets, browsers, and file systems on the user's actual machine. No custom code or API scripting is required.
Meta Ships Llama 3.1 405B as Downloadable Weights
Meta published the full 405 billion parameter Llama 3.1 model under an open license. Developers can download the weights and run inference on local GPUs or rented cloud instances without paying per token fees. The release includes the same tokenizer and chat template used in the hosted version.
RoC AI Skin Insight Turns Selfies Into Quantified Age Scores
RoC AI Skin Insight maps a user selfie to eight skin metrics and returns an estimated skin age plus an overall skin score. The tool uses computer vision trained on dermatological datasets and outputs numeric values for wrinkles, firmness, and tone uniformity. Users receive these scores instantly on the RoC website without downloading an app.
Haut.AI API Lets Brands Ship Custom Skin-Analysis Models
Haut.AI supplies a cloud API that ingests facial images and returns 20-plus skin parameters including porphyrin levels and sebum distribution. Brands integrate the API through REST endpoints and receive JSON responses within 800 milliseconds per image. The company also offers white-label model training so each client receives a version fine-tuned on its own image library.
Sensor Fleets and Sleep Coaches: How Wearables Now Drive Professional Training Decisions
WSC Sports reports that elite teams spend part of the $2.5 billion sports tech market on WHOOP bands, Catapult GPS vests, and hydration patches. These devices record running load, sweat electrolyte loss, and sleep cycles, then feed the numbers into coaching dashboards. Facilities now build nap rooms because the data shows measurable recovery gains.
Budget Sensors Reach the Grassroots: How Schools and Clubs Now Access Former Elite-Only Analytics
embedUR notes that Hudl and Catapult now license cut-down versions of their video and GPS platforms to high-school and club teams. Decathlon sells AI enabled watches under 100 euros that output stride and heart-rate analytics previously limited to pro squads. Programs in India have started using these devices to qualify athletes for national development pathways.
Anthropic Gives Claude Direct Control of Your Desktop
Claude 3.5 Sonnet now uses the new Computer Use API to move the mouse, type on the keyboard, and interact with any desktop application. The model receives screenshots and outputs coordinate-based actions to complete tasks such as filling forms or navigating websites. Anthropic reports the feature reaches 14.9 percent success on OSWorld benchmark tasks without additional fine-tuning.
Meta Hands Over a 405-Billion-Parameter Model for Free
Meta released Llama 3.1 405B under an open license that allows download, fine-tuning, and commercial use without API fees. The model matches or exceeds GPT-4 Turbo on MMLU, HumanEval, and GSM8K benchmarks while running on clusters of eight H100 GPUs. Developers gain full weight access and can host it locally or on any cloud provider.