ChatGPT has now a big problem. Researchers at Oxford and Cambridge exposed a massive threat to large language models.” They call it “model collapse." Internet ecosystem is rapidly changing, and generative AI will soon contribute much of the text found online. This forces us to consider what happens to future iterations like gpt-n when they are trained on data scraped from the web that was already generated by an llm. According to the research, indiscriminately using model-generated content in training causes "irreversible defects" in the resulting ai. the model loses the "tails of the original content distribution." in other words, it begins to forget the creative, fringe, and unique nuances of actual human writing, collapsing into a repetitive echo chamber. This isn't just a chatgpt issue.. the researchers built theoretical intuition showing this collapse is ubiquitous across learned generative models, occurring in large language models as well as in variational autoencoders and gaussian mixture models. Tech companies rely on scraping the internet for large-scale data to build smarter models. However, the paper warns that if we want to sustain the benefits of training on web data, model collapse must be taken seriously. Ultimate takeaway? data collected from genuine human interactions is going to become increasingly valuable in a web filled with ai content.

Everybody is talking about recursive self-improvement (RSI) and meta learning. Here is my old 2020 talk about this [1]. It has aged well. Example: humans still define the starts & ends of trials of many modern meta learners. My RSI systems since 1994 LEARN to (re)define them [2]! [1] Meta Learning Machines in a Single Lifelong Trial (talk for workshops at ICML 2020 and NeurIPS 2021, based on earlier talks since 1994). Abstract: the most widely used machine learning algorithms were designed by humans and thus are hindered by our cognitive biases and limitations. Can we also construct meta learning algorithms that can learn better learning algorithms so that our self-improving AIs have no limits other than those inherited from computability and physics? This question has been a main driver of my research since I wrote a thesis on it in 1987 [2]. Here I summarize our work on meta reinforcement learning with self-modifying policies in a single lifelong trial (since 1994), and mathematically optimal meta-learning through the self-referential Gödel Machine (since 2003). Many additional publications on meta-learning since 1987 can be found in the RSI overview [2]. [2] J. Schmidhuber (AI Blog, 2020-2025). 1/3 century anniversary of first publication on recursive self-improvement (RSI) and meta learning machines that learn to learn (1987). For its cover I drew a robot that bootstraps itself. 1992-: gradient descent-based neural meta learning. 1994-: meta reinforcement learning with self-modifying policies. 1997: meta RL plus artificial curiosity and intrinsic motivation. 2002-: asymptotically optimal meta learning for curriculum learning. 2003-: mathematically optimal Gödel Machine. 2020-: new stuff!

Today everyone is talking about Recursive Self-Improvement (RSI). In 1987, when compute was 100,000,000 x more expensive, I published the 1st concrete RSI algorithms. Now compute is cheap, and RSI is driving the future of both software and physical AI. See: RSI since 1987 people.idsia.ch/~juergen/rec… (Technical Note IDSIA-9-26) Also covered: RSI with self-modifying policies since 1994, gradient descent-based RSI in neural networks since 1992, asymptotically optimal RSI for curriculum learning since 2002, mathematically optimal RSI through the self-referential Gödel Machine since 2003, RSI combined with artificial curiosity and intrinsic motivation since 1990/1997, recent work on RSI since 2020. Software-based RSI has become practical. Full RSI, however, will require not just self-improving software but self-improving hardware in the physical world. As of 2026, companies talking about RSI include Anthropic, OpenAI, Sakana AI, SpaceX, Ricursive, Recursive Superintelligence, Inherent …

Google Quantum AI Researchers, This proposal for using Google's Quantum AI capabilities to develop advanced magnetocorticographic neural implants is from Gemini (and me): Google Quantum AI, "Quantum is the language of nature," and as Richard Feynman noted, to make a simulation of nature, we'd better make it quantum mechanical. Simulating the natural world requires computing systems that inherently operate on quantum principles, and Google Quantum AI aims to build quantum computing for these otherwise unsolvable problems. We have the ultimate unsolvable challenge ready for you: Magnetocorticographic Neural Implants. Here is why Google's Quantum capabilities are the missing key: - Why Quantum is Needed For The Material Science & Physics: Developing the ultra-sensitive components of this implant relies heavily on quantum phenomena, such as spin-dependent tunnelling and electrostatic spin-orbit modulation. To design the perfect magnetic interface, molecules, materials, and living systems must be studied as they truly behave. Achieving this atomic-level simulation is an otherwise unsolvable problem for classical computers. - The Challenges Are Already Organized For You: The material science roadmap is specifically mapped out for Google's exact quantum solvers, which are needed to compute ferromagnet-oxide interfaces exactly before anything is physically grown. We also require your systems to run spin-aware world models to simulate the deposition and annealing processes, allowing us to read off exactly where every single atom ends up. - Monumental Neuroscience & Neurology Outcomes: Solving these material challenges will allow us to discover the true neural correlates of consciousness and watch a memory form, consolidate, and be recalled. This technology will turn neurology into a measurement-based precision science. It will enable closed-loop therapy for the entire cortex, cure epilepsy, restore movement for those with paralysis, and detect neurodegeneration decades before symptoms appear. - Tremendous Benefits For Future AI Evolution At Google: A successful quantum-developed implant will produce pristine, whole-cortex data that is necessary to build foundation models of the human cortex. This flawless, vector-based dataset will redirect material searches to further sharpen our image of thought. By providing an unbroken biological ground truth, this neural data will force AI to evolve into highly efficient, neuromorphic designs that mirror actual human cortical dynamics—propelling Google directly to true AGI. See the attached "Design Of Future Magnetocoticographic Neural Implants Part 1.png" and "Design Of Future Magnetocoticographic Neural Implants Part 2.png" for the complete physics blueprints and AI roadmaps. Let's use quantum computing to unlock the human brain!

The holy grail of RSI is the mathematically optimal self-referential Gödel Machine (2003) [GM3-9] (Sec. 5 of the RSI report above). It inspired lots of recent work, e.g., the Darwin-Gödel Machine (2025) [GMD25], the Huxley-Gödel Machine (2026) [GMH26], the Red Queen Gödel Machine (2026) [GMR26]... See also our 2026 survey [RSI26] and its tweet x.lingyaoai.com/SchmidhuberAI/status/2…