🐸 Harvard Creates Cyborg Tadpoles With Brain Implants (Because Regular Tadpoles Were Too Mainstream)

Remember when AI was just for making weird art and cheating on homework? Well, turns out it's actually decent at discovering drugs that don't kill people. This week, we're diving into AI's surprising pharmaceutical success rate, China's regulatory glow-up that's making the FDA sweat, apparently your brain is also in your gut, and tadpoles with brain implants (because regular tadpoles were apparently too boring). Oh, and Silicon Valley wants to eliminate the need for human eggs in reproduction. Because, of course, they do. Let's unpack this beautiful chaos!

While tech bros were debating whether AI would achieve consciousness, it was quietly getting good at something actually useful: drug discovery. By 2023, 67 AI-discovered molecules made it to clinical trials, up from basically zero a decade ago. That's exponential growth that would make even crypto bros jealous.

Here's the spicy part: these AI drugs are showing 80-90% success rates in Phase I trials, compared to the traditional pharma average of 40-70%. Before you liquidate your 401k to invest in AI biotech, remember this only covers 21 drugs that completed Phase I, basically the "don't immediately kill people" phase. Phase II success rates drop back to ~40%, matching regular drugs. So AI is great at picking non-toxic molecules, but still struggles with the "actually works" part.

The criteria for "AI-native" drug discovery are hilariously vague. Companies qualify if they use AI for at least one of five things: target discovery, drug repurposing, small molecule design, biologics optimization, or vaccine design. That's like calling yourself a chef because you once used a microwave. Boston Consulting Group, which compiled the data, basically admitted that anything fancier than Excel counts as AI these days.

What's impressive is the speed. Traditional drug discovery takes 10-15 years and costs $2.6 billion per approved drug. AI companies are claiming they can cut both in half. Recursion Pharmaceuticals is testing 2.6 million compounds per week using robots and AI. Atomwise screened 10 million compounds for COVID drugs in days, not years. Even if half of these companies are just riding the hype train, the other half might actually revolutionize how we find medicines.

Five years from now, the entire drug pipeline might look completely different. Or we'll discover AI was just really good at finding molecules that taste like chicken. Time will tell.

You can read more from here and here

China just pulled a regulatory power move that has Big Pharma adjusting its flight schedules to Beijing. The NMPA proposed slashing clinical trial review times from 60 to 30 working days, matching FDA timelines. This is for the good stuff: pediatric cancer drugs and rare disease treatments. When China decides to move fast, apparently even its legendary bureaucracy gets out of the way.

Why should you care? Because last year, large pharma licensed 31% of their external molecules from China, compared with 10-12% in 2020. That's not a typo. China went from a biotech side character to the main protagonist in four years. Companies like BeiGene and Innovent are no longer just copycats; they're creating novel drugs that Western pharma desperately wants.

The numbers are staggering. China's pharmaceutical market hit $193 billion in 2024 and is projected to reach $320 billion by 2030. They're graduating 350,000 life science PhDs annually (the US manages 65,000). The US Congress is so spooked they're asking for $15 billion over five years just to keep up, which in government spending terms is basically pocket change.

The shift is real: Chinese biotechs aren't just making generic aspirin anymore. They're innovating faster than Silicon Valley pivots to AI, and now they're streamlining approvals to match. BeiGene's tislelizumab beat Western PD-1 inhibitors in head-to-head trials. Zai Lab's tumor-treating fields device got FDA approval before anywhere else. When bureaucracy moves this fast, you know someone's serious about winning.

The geopolitical implications are wild. The BIOSECURE Act is trying to block US companies from working with Chinese biotechs, but good luck putting that genie back in the bottle when a third of new drugs are coming from China. Pro tip for pharma execs: Start learning Mandarin. Your next blockbuster drug might come with assembly instructions in Chinese.

Scientists at Harvard decided regular tadpoles were too mainstream, so they created cyborg ones with neural implants that grow WITH the brain. Published in Nature, this isn't your grandfather's brain chip. It's more like your brain's new roommate that moved in during embryonic development and never left.

The genius/terrifying part: they implant the electronics when the tadpole is just an embryo. As the brain develops, the implant becomes part of the neural tissue, like a technological birthmark, but for thoughts. The tadpoles developed normally, suggesting the brain just shrugged and said, "Guess we're cyborg now." The mesh electronics stretch and grow with the brain tissue, maintaining stable recordings for the entire developmental period.

This solves a huge problem in neuroscience. Traditional rigid implants cause scarring and get rejected faster than a bad Tinder date. The brain treats them like the foreign invaders they are, surrounding them with scar tissue until they stop working. These soft, flexible arrays integrate so well that the brain doesn't even notice it's been upgraded to Tadpole 2.0. The immune system basically ghosted the implants entirely.

Lead researcher Jia Liu's team used a polymer mesh that's 1,000 times more flexible than conventional electronics. They inject it through a needle thinner than a human hair, and it unfurls inside the brain like those foam dinosaurs you put in water as a kid. The mesh is so biocompatible that neurons actually grow through it, creating a true brain-machine hybrid that would make the Borg jealous.

Next step: mammals. Then humans. Then we're all walking around with built-in WiFi, wondering why we ever thought Neuralink's approach made sense. The future is soft, squishy, and slightly disturbing. But at least our brain implants won't try to escape like angry porcupines.

Read more here or read the Nature article

Conception Bio, backed by OpenAI's Sam Altman, is trying to turn any cell into a human egg. Yes, you read that right. They want to take your blood cell and convince it to become an egg. It's like a cellular identity crisis as a service. The company raised $20 million to essentially gaslight skin cells into thinking they're reproductive material.

The science is called in vitro gametogenesis (IVG), which sounds like something you'd order at a fertility clinic's coffee shop. The process involves taking adult cells, reverting them to stem cells, then coaxing them to become eggs or sperm. Japanese scientists already did this with mice in 2016, producing healthy pups from tail cells. Now Silicon Valley wants to scale it up to humans because, of course, they do.

The company admits they "haven't done it yet, nor has anyone else", which is refreshingly honest for a startup. But the implications are wild: same-sex couples could have biological children sharing both parents' DNA, women could have kids at 70 without donor eggs, and fertility clinics would become obsolete. Post-menopausal pregnancy could become as common as post-menopausal hot flashes.

The ethical minefield is spectacular. We're talking about creating human life from nail clippings. Religious groups are already sharpening their pitchforks. Bioethicists are speed-writing papers about consent, identity, and whether your discarded hair follicles have reproductive rights. The FDA doesn't even have a regulatory framework for this yet; they're basically making it up as they go.

The timing is suspicious. Chinese researchers had recently created mouse offspring from two fathers. The UK is considering legalizing three-parent babies. Japan's population crisis has them throwing money at any reproduction technology that moves. If this works, the culture wars are about to get a whole new chapter titled "But What About The Cellular Parents?"

Altman's involvement makes sense. After trying to create artificial general intelligence, artificial human reproduction seems like a logical next step. Though explaining to your kid that they were conceived by venture capital might require some creative parenting. "Well, honey, when a VC and a startup love each other very much, they create a term sheet..."

Scientists at Duke-NUS just caught your intestinal cells red-handed using neuron-like tentacles to whisper sweet nothings to each other. Turns out your gut has been running its own cellular WhatsApp group for years, and we're just now figuring out the group chat.

Your intestinal lining replaces itself every 4-5 days, which is basically like renovating your entire house twice a week, forever. Scientists thought this happened through simple chemical diffusion, like spraying air freshener and hoping for the best. Wrong. Telocytes (Greek for "perfect cells," because apparently even cells have egos) are extending incredibly long cellular arms called cytonemes to make direct contact with intestinal stem cells.

Dr. Gediminas Greicius put it perfectly: "This system of targeted signaling was hiding in plain sight." His colleague Professor David Virshup added that these signals aren't just drifting through tissue but are "being delivered with surprising precision... similar to how neurons pass signals to one another in the brain".

Here's where it gets spicy: when researchers deleted telocytes' ability to produce Wnt signals (the molecular equivalent of cutting their phone lines), intestinal stem cell renewal completely stopped within 24-72 hours. That's right, these cells aren't just helpful, they're the difference between having a functioning gut and, well, not.

The implications are huge. About 80% of colorectal cancers involve mutations in this exact signaling pathway. Inflammatory bowel diseases like Crohn's? Same story. If we can hack into this cellular communication network, we might finally have realistic treatments instead of just managing symptoms. The timeline looks surprisingly achievable: biomarkers in 2-5 years, targeted therapies in 5-10 years.

Of course, there's scientific drama. Some researchers think telocytes are just fancy fibroblasts with a marketing problem. Plus, the original discoverer turned out to be a Romanian secret police informant, because even cellular biology needs a spy thriller subplot. But the evidence keeps mounting, and your gut's secret neural network isn't going anywhere.

You can also read more here and here

Another week, another reminder that science fiction writers need to step up their game… again. From AI actually getting actually more useful in pharma to tadpoles getting designer brain implants, reality is getting weird faster than we can regulate it.

Did the cyborg tadpoles creep you out? Ready to move your biotech startup to China for faster approvals? Planning to wait for artificial eggs before having kids? Hit reply and share your thoughts. We read every email while checking if our tadpoles have achieved consciousness!

Share this with someone who thinks science moves too slowly. We're at 2,000+ subscribers and growing faster than China's biotech sector!

Stay curious, stay skeptical, and maybe check if your tadpoles have any unusual electronics, Prateek & Jere

And if you’re still not convinced, no worries! Just unsubscribe here. You can always check back on us later.