Could computers of the future run on human brain cells? A team of researchers at Johns Hopkins University certainly think so. In a paper published in the journal Frontiers in Science, the team outline their plans for ‘organoid intelligence’, an emerging multidisciplinary field looking to develop biocomputers that operate with human brains cells. Such a development could not only massively expand the capabilities of modern computing but also open up new fields of study.
Organoids are tiny, self-organizing 3D tissues that are typically derived from stem cells, and mimic the main functional and architectural complexity of an organ. It is possible there could be as many types of organoids as there are tissues and organs in the body. To date, scientists have produced organoid cultures for intestines, liver, pancreas, kidneys, prostate, lung, optic cup, and the brain, and it seems more may be on the way.
These tissues provide unique opportunities for scientists to study human diseases that do not rely on traditional methods associated with animal models. The reliance on animal models has historically led to a bottleneck in treatment discovery as there are biological processes that are specific to the human body and cannot be modeled on animals. The development of organoids promises to overcome these limitations. Yet the team at Johns Hopkins University are taking the research into organoids in a completely different direction.
“Computing and artificial intelligence have been driving the technology revolution but they are reaching a ceiling,” explained Thomas Hartung, a professor of environmental health sciences at the Johns Hopkins Bloomberg School of Public Health and Whiting School of Engineering, in a statement. “Biocomputing is an enormous effort of compacting computational power and increasing its efficiency to push past our current technological limits.”
We’ve trained a model called ChatGPT which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests. ChatGPT is a sibling model to InstructGPT, which is trained to follow an instruction in a prompt and provide a detailed response.
We are excited to introduce ChatGPT to get users’ feedback and learn about its strengths and weaknesses. During the research preview, usage of ChatGPT is free. Try it now at chat.openai.com
We trained this model using Reinforcement Learning from Human Feedback (RLHF), using the same methods as InstructGPT, but with slight differences in the data collection setup. We trained an initial model using supervised fine-tuning: human AI trainers provided conversations in which they played both sides—the user and an AI assistant. We gave the trainers access to model-written suggestions to help them compose their responses.
To create a reward model for reinforcement learning, we needed to collect comparison data, which consisted of two or more model responses ranked by quality. To collect this data, we took conversations that AI trainers had with the chatbot. We randomly selected a model-written message, sampled several alternative completions, and had AI trainers rank them. Using these reward models, we can fine-tune the model using Proximal Policy Optimization. We performed several iterations of this process.
ChatGPT is fine-tuned from a model in the GPT-3.5 series, which finished training in early 2022. You can learn more about the 3.5 series here. ChatGPT and GPT 3.5 were trained on an Azure AI supercomputing infrastructure.
ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers. Fixing this issue is challenging, as: (1) during RL training, there’s currently no source of truth; (2) training the model to be more cautious causes it to decline questions that it can answer correctly; and (3) supervised training misleads the model because the ideal answer depends on what the model knows, rather than what the human demonstrator knows.
ChatGPT is sensitive to tweaks to the input phrasing or attempting the same prompt multiple times. For example, given one phrasing of a question, the model can claim to not know the answer, but given a slight rephrase, can answer correctly.
The model is often excessively verbose and overuses certain phrases, such as restating that it’s a language model trained by OpenAI. These issues arise from biases in the training data (trainers prefer longer answers that look more comprehensive) and well-known over-optimization issues.12
Ideally, the model would ask clarifying questions when the user provided an ambiguous query. Instead, our current models usually guess what the user intended.
While we’ve made efforts to make the model refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behavior. We’re using the Moderation API to warn or block certain types of unsafe content, but we expect it to have some false negatives and positives for now. We’re eager to collect user feedback to aid our ongoing work to improve this system.
Today’s research release of ChatGPT is the latest step in OpenAI’s iterative deployment of increasingly safe and useful AI systems. Many lessons from deployment of earlier models like GPT-3 and Codex have informed the safety mitigations in place for this release, including substantial reductions in harmful and untruthful outputs achieved by the use of reinforcement learning from human feedback (RLHF).
Multivariate Time Series Forecasting focuses on the prediction of future values based on historical context. State-of-the-art sequence-to-sequence models rely on neural attention between timesteps, which allows for temporal learning but fails to consider distinct spatial relationships between variables. In contrast, methods based on graph neural networks explicitly model variable relationships. However, these methods often rely on predefined graphs and perform separate spatial and temporal updates without establishing direct connections between each variable at every timestep. This paper addresses these problems by translating multivariate forecasting into a spatiotemporal sequence formulation where each Transformer input token represents the value of a single variable at a given time. Long-Range Transformers can then learn interactions between space, time, and value information jointly along this extended sequence. Our method, which we call Spacetimeformer, achieves competitive results on benchmarks from traffic forecasting to electricity demand and weather prediction while learning fully-connected spatiotemporal relationships purely from data.
This is a project that ressembles the classic Game of Life, but implemented in a 3D world. The project needs OpenGL 4.5 or greater to render, and currently only compiles under Linux (although it would be relatively easy to change CMakeLists to compile for Windows/macOS*(apple discontinued OpenGL support in favor of Metal)).