Microsoft AI Path to Medical Superintelligence Microsoft is set out to leverage AI in the medical field to support diagnosis with their latest 304-case sequential-diagnosis benchmark for LLMs: It is quite interesting to see the intersection of AI in key sectors like healthcare, with Microsoft releasing a benchmark that forces AI models to decide which questions and tests to order (each with a simulated dollar price) before making a diagnosis. This new Medical AI Diagnostic Orchestrator proposes wrapping any frontier LLM in a "mixture-of-experts" agent architecture that self-checks reasoning and enforces cost ceilings. The provided benchmarks were tested with OpenAI o3 and claimed 85.5% accuracy (vs 20% for practicing clinicians), all while spending less on tests than either humans or standalone models. These type of initiatives have to certainly be taken with a level of skepticism given the hype, but it is certainly great to see the investments from research labs to tackle some of these important topics, and particularly in the case of open frameworks / models. |
|
|
|---|
|
OpenAI Practical Guide to Agents OpenAI has released a comprehensive guide to building agents to empower developers when creating AI-enabled applications beyond chatbots: This guide from the OpenAI team focuses on applications that encompass entire workflows that making complex decisions, invoke tools, and relinquish control only on failure. It is interesting to see the differences between orgs publishing these guides; OpenAI is (expectedly) bullish on deploying agentic workflows in contexts that requrie messy unstructured data whilst leaving straightforward tasks to simpler data processing pipelines. Most of the topics from these type of guides are starting to converge towards a generic best practice, enabling for a more standardised set of guidelines to emerge that ideally can become cross-cutting beyond just a single AI lab. |
|
|
|---|
|
McKinsey's Trillion War for Data Centers Compute is the new oil; McKinsey forecasts that global data-centre capacity must almost triple by 2030, demanding ~$7 trillion in new capital: This recent report from McKinsey provides important insights for AI leaders, quantifying that AI workloads driving about 70% of the forecasted ~$7Tn growth ($5.2 trillion for AI alone); some scenarios push the forecast as high as ~$8 trillion. About 60% of this spend seems will go to chips and servers, 25% to power and cooling infrastructure, and 15% to land and construction. This will be leaving silicon vendors, utilities and builders in a high-stakes race where capacity, energy and supply-chain constraints (not algorithmic breakthroughs). Similarly there will be a lot of important considerations in regards to sustainability and efficiency as these demands grow in order to ensure sustainable growth globally. |
|
|---|
|
Bloom Filters by Example Probabilistic data structures such as Bloom Filters are a key concept in distributed systems that we interact with on a daily basis, however there's huge value to understand the internals: This is a great overview on the internals of Bloom filters that provides visual / interactive examples on how this data structure works as data is provided / fetched. A Bloom filter is a probabilistic data structure that answers “have I seen x?” in O(k) time whilst using reduced memory through probabilistic approximations. Namely, Bloom filters sometimes say "maybe" when the answer is actually "no" in order to ensure memory efficiencies, however this data structure never misses a true hit. These probabilistic tradeoffs enable slashing storage and compute costs across data deduplication, spam blocking, database queries, and real-time recommendation feeds. This is one of the best intuitive overviews of Bloom Filters out there, so it's definitely recommended to check it out in detail! |
|
|
|---|
|
The Jax ML Scaling Book As of today, scale in ML tends to define success across some of the leading players in the space - Google's "Scaling Book" is one of the best resources for this: There are really great tips in this scaling book, covering key topics in ML scaling such as rooflines, TPUs, Sharded Matrix Multiplications, provides an overview on transformers, as well as nuances on training such as GPU utilisation, and dives into practical cases such as training LLaMA, inference and profiling. |
|
|
|---|
|
Upcoming MLOps Events The MLOps ecosystem continues to grow at break-neck speeds, making it ever harder for us as practitioners to stay up to date with relevant developments. A fantsatic way to keep on-top of relevant resources is through the great community and events that the MLOps and Production ML ecosystem offers. This is the reason why we have started curating a list of upcoming events in the space, which are outlined below. Upcoming conferences where we're speaking:
Other upcoming MLOps conferences in 2025:
In case you missed our talks:
|
|
|---|
| | |
Check out the fast-growing ecosystem of production ML tools & frameworks at the github repository which has reached over 10,000 ⭐ github stars. We are currently looking for more libraries to add - if you know of any that are not listed, please let us know or feel free to add a PR. Four featured libraries in the GPU acceleration space are outlined below. - Kompute - Blazing fast, lightweight and mobile phone-enabled GPU compute framework optimized for advanced data processing usecases.
- CuPy - An implementation of NumPy-compatible multi-dimensional array on CUDA. CuPy consists of the core multi-dimensional array class, cupy.ndarray, and many functions on it.
- Jax - Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
- CuDF - Built based on the Apache Arrow columnar memory format, cuDF is a GPU DataFrame library for loading, joining, aggregating, filtering, and otherwise manipulating data.
If you know of any open source and open community events that are not listed do give us a heads up so we can add them! |
|
|---|
| | |
As AI systems become more prevalent in society, we face bigger and tougher societal challenges. We have seen a large number of resources that aim to takle these challenges in the form of AI Guidelines, Principles, Ethics Frameworks, etc, however there are so many resources it is hard to navigate. Because of this we started an Open Source initiative that aims to map the ecosystem to make it simpler to navigate. You can find multiple principles in the repo - some examples include the following: - MLSecOps Top 10 Vulnerabilities - This is an initiative that aims to further the field of machine learning security by identifying the top 10 most common vulnerabiliites in the machine learning lifecycle as well as best practices.
- AI & Machine Learning 8 principles for Responsible ML - The Institute for Ethical AI & Machine Learning has put together 8 principles for responsible machine learning that are to be adopted by individuals and delivery teams designing, building and operating machine learning systems.
- An Evaluation of Guidelines - The Ethics of Ethics; A research paper that analyses multiple Ethics principles.
- ACM's Code of Ethics and Professional Conduct - This is the code of ethics that has been put together in 1992 by the Association for Computer Machinery and updated in 2018.
If you know of any guidelines that are not in the "Awesome AI Guidelines" list, please do give us a heads up or feel free to add a pull request!
|
|
|---|
| | |
| | | | The Institute for Ethical AI & Machine Learning is a European research centre that carries out world-class research into responsible machine learning. | | | | |
|
|
|---|
|
|
This email was sent to You received this email because you are registered with The Institute for Ethical AI & Machine Learning's newsletter "The Machine Learning Engineer"
|
| | | | |
|
|
|---|
|
© 2023 The Institute for Ethical AI & Machine Learning |
|
|---|
|
|
|
The Institute for Ethical AI & Machine Learning
