Brainify - AI & Future Tech

Quick introduction. I'm an atheist in Sydney who reads clinical trials for fun, which my friends find either inspiring or deeply worrying. I'm a hobbyist, not a scientist, but I love science because it's the only method we have that admits when it's wrong and then fixes itself. In 4,000 years of prayer, nobody has regrown a limb. In about 40 years of medical research, we've got gene therapy giving deaf kids their hearing, vaccines that wiped out smallpox, and implants that let paralysed people move a robot arm with their thoughts. Draw your own conclusions. I already have. So this account is my small war on the people who demonise science: the anti-vax uncle, the 'doctors don't want you to know' salesman, the preacher who says the earth is 6,000 years old while recording on a smartphone built on quantum physics. Every day I take a real study, check who ran it and whether it holds up, and explain it in plain English and Arabic. If you've got evidence I'm wrong, bring it. If you've only got feelings, bring popcorn.

Researchers at Griffith University on the Gold Coast have been teaching people with complete spinal cord injuries to walk in virtual reality using only their thoughts, and the brain gets better at it with practice. The setup is a consumer-grade EEG headset with dry electrodes, linked to a VR headset. You imagine taking a step, the system reads that pattern in your brain waves, and your virtual legs walk. The visuals and sounds match what you imagined, so your brain gets feedback it hasn't had since the injury. Over several sessions, five people without injuries learned to control their brain signals more reliably. Four people living with chronic complete spinal cord injury showed the same improvement, and they did it in real rehab clinics rather than a tidy lab. That matters, because after a spinal injury the brain slowly loses its map of how movement feels. This kind of training looks like a way to keep that map alive and sharpen it. It's a small study and nobody stood up from a wheelchair. What it does show is a practical tool built from consumer-grade gear that could sit alongside spinal stimulation and other therapies, giving the brain something useful to do while the body catches up. Miracles have a famously poor track record with spinal cords. Engineers on the Gold Coast with a VR headset seem to be doing rather better. العلم يفتح الأبواب Paper in Scientific Reports: doi.org/10.1038/s41598-026-3…

If you pick up a nasty infection in a country town in Western Australia, the doctor treating you may be a long way from the nearest infectious disease specialist. Choosing the right antibiotic for bugs that may already be resistant in that region is a hard call to make alone late at night. An Australian company, Kraken Coding, built a decision support tool called Clinical Branches that turns the therapeutic guidelines into step by step advice. It takes in the patient's details and the local resistance patterns, so the guidance fits the town you're working in. Researchers at UTS, working with St Vincent's Hospital Sydney, independently evaluated it in WA Country Health Service hospitals. Where staff took it up, prescribing that matched clinical guidelines improved by nearly 21 per cent. That matters because new antibiotics are arriving far more slowly than resistant bacteria are evolving. Dr Sarah Wise from UTS calls antimicrobial resistance the sleeping giant of public health, and says our best defence right now is using the drugs we already have wisely. Rural and remote communities, including many Aboriginal and Torres Strait Islander communities, carry more of that risk. The evaluation was frank about the weak spots too. Plenty of staff had never heard of the tool, and high turnover meant people moved on before they'd learned it. The clinicians who did use it trusted it, so the next job is showing every new doctor and nurse in a remote clinic how it works from their first week. Bacteria have been dodging our medicines since the 1940s without any help from above, so it's good to see us putting some intelligence of our own into keeping up. Would you trust an AI tool helping your doctor choose your treatment? uts.edu.au/news/2026/10/ai-g…

Anthropic just filed to go public and headlines are screaming about an $80.7 billion loss. Read past paragraph one and $72 billion of that is non-cash accounting noise: convertible notes revaluing upward because the company got more valuable, not less. Strip that out and the real operating loss is $693 million against $16.3 billion in revenue that grew over 1,300% in six months. That's not a company bleeding out, that's a filing nobody bothered to read past the headline. 🧠 #AI #Anthropic

A 2mm chip just gave a Stargardt disease patient his vision back, one letter at a time. Science Corporation implanted it at Sydney Eye Hospital in May, and by his second rehab session he was reading whole words again. Stargardt disease eats away at central vision in roughly 150,000 people worldwide and has no cure. What the implant gives back isn't full sight, it's small, sharp, black-and-white, like looking through a straw. For someone who had lost the ability to read, that straw is everything. #Medicine #Accessibility

Richard Dawkins is bringing The Selfish Gene to Sydney Town Hall on November 10, fifty years after he explained the single idea that still makes people furious: you're not driving the bus, your genes are, and your happiness was never the destination. It wasn't nihilism, it was honesty, and honesty about biology still scares people more than any sermon does. 🧬 #Atheism #CriticalThinking

Four people just finished a 378-day fake Mars mission inside a 3D-printed box in Houston, and the hardest part wasn't the isolation. NASA built in the same communication delay a real Mars crew would face talking to Earth, on purpose. They ran simulated spacewalks, grew their own food, and kept the habitat running while quietly generating the psychological data NASA actually needs before sending anyone to Mars for real. The actual mission is still years out. These four just did the slow, unglamorous version first, and stepped outside for the first time in over a year. 🛰️ #Space #NASA

Lung cancer is Australia's deadliest cancer, and the reason it's so deadly is mostly timing. By the time it causes symptoms, it's usually well advanced. Catching small nodules early, and treating them fast, is where lives get saved. The trouble is that the smallest nodules tend to sit deep in the outer branches of the lungs, where a normal scope struggles to reach. Respiratory specialists at Macquarie University Hospital in Sydney, Professors Alvin Ing, Tajalli Saghaie and Jonathan Williamson, have been steering robotic bronchoscopes through the airways to get there. Since 2022 they've led first-in-human trials that used the robot to biopsy those hard-to-reach spots. Now they're using it for treatment as well. With a newly installed Ion robotic platform, the team places tiny fluorescent coils inside faint, hazy "ground glass" nodules so surgeons can find and remove them precisely with keyhole robots. Professor Saghaie is also leading the CONVERGE trial for people with Stage III non-small cell lung cancer, where the robot injects nanoparticles containing hafnium, a heavy metal, straight into the tumour to make radiation therapy hit harder. Macquarie has also opened a new research lab, MQ BRIL, which builds realistic models of human airways so new tools can be tested and doctors can train on them before they reach a patient. They want Sydney to become a training hub for the whole Asia-Pacific region. I find it quietly wonderful that a robot can now thread its way through the tiny branching tubes of a human lung and drop a glowing marker exactly where the surgeon needs it. Our lungs ended up as this tangled maze because evolution never planned ahead, and it's good to see clever people at Macquarie finding a way through it. lighthouse.mq.edu.au/article…

OpenAI says its AI disproved a 25-year-old assumption underpinning huge chunks of complexity theory, alongside a decades-old harmonic analysis conjecture and a quantum field theory result tied to the Yang-Mills Millennium Prize problem. The maths went straight to GitHub with a full formal proof, same day, for anyone to check. The physics went to a hand-picked panel that decided what the rest of us get to see, and when. OpenAI itself admits that setup creates an 'asymmetry of information.' A real breakthrough survives daylight. One that needs a gatekeeper doesn't. 🧠 #AI

Your eyes and brain handle a huge stream of visual information on remarkably little energy, and they're very good at noticing what has changed in a room. Engineers at RMIT in Melbourne have been trying to copy that trick, and they now have a working prototype. Professor Sumeet Walia's team at the Centre for Opto-electronic Materials and Sensors built a 2cm by 2cm chip from molybdenum disulfide, a semiconductor only one atom thick. The chip senses light, remembers what it saw and does the processing right there, so it doesn't have to send every frame off to a separate computer. In lab tests it learned to recognise numbers, shapes and movement, and it picks up when something in the scene changes. That matters for anyone hoping for a better bionic eye. A future device would have to work out what changed, keep the useful information and send signals quickly on a tiny power budget, probably sitting close to the body. Doing the sensing and the thinking in one place is a real step towards that. The team also came up with a cleaner, water-based way to make these atom-thin devices with fewer defects. Manufacturing wins like that rarely make headlines, but they decide whether a lab idea ever reaches people. RMIT has filed an international patent. Walia is upfront that a working bionic eye is still many years and many breakthroughs away, and the first uses will probably be in robots and drones. I appreciate that honesty. Evolution took hundreds of millions of years of trial and error to build the eye with nobody at the drawing board, so a Melbourne lab making progress on copying it is pretty impressive. rmit.edu.au/news/all-news/20…

AI is making it much faster to design new drugs, and biotech investor Bruce Booth points out a catch. Most of that speed is going into well known targets like PD-1, EGFR and KRAS, and about a quarter of all drug programs already chase just 37 targets. Lots of near copies chasing the same patients slows trials for everyone. I'd love to see that speed aimed at rare diseases and disabilities with no treatment at all. Where would you point it?

Photos of a Kenyan boy with cerebral palsy, sitting in one shot and walking in the next, are being shared as proof of a prophet's miracle. No independent medical check has been made public, and Kenya's medical regulators warn that claims like this can lead families to skip treatment that works. Physio, surgery and good assistive gear have changed countless lives with cerebral palsy, and their results can be checked by anyone. Miracles rarely seem to arrive with a scan attached. What evidence would you want to see?

Researchers in Japan found two ordinary bacteria, one taken from tumour tissue and one from soil, that cleared every tumour in mice with colorectal, ovarian and pancreatic cancers after two IV doses. Antibiotics clear the bacteria out afterwards. Mouse results often don't hold up in people, and human trials are about six years off, so patience is needed. Which cancer would you most want this tested on first?

A small startup has made glasses that pick up the tiny muscle movements you make when you silently mouth words and turn them into text at over 200 words per minute. They use low power radio waves instead of a camera or microphone. I keep thinking about people who have lost their voice but can still move their mouth. Could this turn out more useful as assistive tech than as a gadget for texting?

A totally blind man took his first ride in a Tesla Robotaxi and called it unusually normal. The car honked so he could find it, and voice commands handled everything else. Tesla also asked wheelchair users and other disabled riders to help shape the Cybercab, and Braille controls and grab handles came out of it. For a lot of blind people, getting around alone is the hardest part of the day. Would you trust a driverless car with your commute?

UNSW Sydney is about to train the next generation of grid-battery engineers, and the company funding it is Chinese. CATL just unveiled a renewable energy training center in Western Sydney, fresh off launching a storage system that crams 9 megawatt-hours into 40% fewer containers and is built to run for 25 years. This is the same company that already controls 27% of the global battery storage market, more than double its nearest rival. Australia isn't debating whether grid-scale storage is real. It's letting a Chinese company train the workforce that will run it. ⚡ #CleanEnergy #Australia

Chemotherapy still involves a fair bit of educated guessing. An oncologist picks a drug, the patient goes through months of side effects, and sometimes the cancer had already worked out how to survive it. A team at ANU in Canberra wants to take the guessing out. Dr Eli Niktab, Principal AI Engineer at the National Centre for Indigenous Genomics, is leading a project that has just won about $3.1 million (US$2.2 million) from the Google.org Impact Challenge: AI for Science. ANU was one of 17 groups picked from more than 3,000 applicants in 30 countries. They're building a Cytotoxic Resistance Atlas, an open molecular map of cancer cell lines known to survive chemotherapy, plus an AI engine that can spot the genetic tricks a tumour uses to resist treatment, using samples taken during routine care. Partners include Canberra Hospital, the Victorian Cancer Biobank and the University of Kent's resistant cell line collection, and the models will train on Gadi, Australia's national supercomputer. The part I like most is who it's built for. Existing tests were developed almost entirely on data from people of European ancestry, so Aboriginal and Torres Strait Islander Australians, and plenty of others, get the least reliable answers. This model learns from the full diversity of human genomes, and the atlas, models and software will all be free for researchers to use. Cancer cells have been evolving their way around our drugs for as long as we've had drugs, which is an odd thing for an intelligent designer to sign off on. Good to see a few humans with a supercomputer catching up. jcsmr.anu.edu.au/news-events…

Anyone who has done rehab knows the hardest part is often the boredom. You repeat the same movement hundreds of times, and motivation drains away long before the progress shows up. Two UNSW engineering students, Paige Tinyow and Raymond Liu, decided to do something about that. Their system, Rehabit, fits custom handles with silicone grips onto the handbikes already in the rehab gym at Prince of Wales Hospital in Randwick and links them to interactive games, so the exercise turns into play. Raymond designed the hardware, and Paige wrote the software and built the games. It's already in use in that gym, and clinicians are collecting data on how patients interact with it and how well it performs. The project has just won $20,000 through the UNSW Founders Peter Farrell Cup, and the money goes towards preparing for manufacturing, ethics approval and clinical trials. They want it in hospitals right across Australia. I love this kind of science because it's cheap and practical, and it was designed alongside the physios who'll actually use it. People have prayed for faster recoveries for a few thousand years without much to show for it, so it's lovely to see someone try a handbike and a scoreboard instead. unsw.edu.au/news/2026/10/tyr…

The Digital Duty of Care Bill passed the House of Representatives 96 to 38. It would make social media, games and AI chatbots assess risks and protect kids, and Senate hearings start in early November. Supporters call it overdue. Critics worry about surveillance and vague definitions of harm. Where do you land on it?

Surgeons in Russia say they've done the first spinal cord transplant in people. New Scientist says the evidence behind it is thin. I'd love spinal repair to work, and that's why I want to see the data before I celebrate. What would you need to see before you believed it?

A man with ALS is talking at 75 words per minute again using only a Neuralink implant, with a 5.8% error rate in his first week. The team spent 72 hours at his house so he could make phone calls with it. Eye gaze boards are painfully slow compared to that. Who in your life would this change things for?