Andrew Ng on Building Faster This ~40min keynote from Andrew Ng at Y-Combinator is one of the best overviews for the state of rapid AI prototyping: Andrew brings some of the key learnings from ML startups leveraging agentic workflows and AI coding assistants to spin up disposable prototypes. The main takeaways are emphasising what most of us have seen in industry, namely that tooling is enabling us to treat architecture choices as reversible “two-way doors”, and sharpening direction through fast feedback loops. As engineering accelerates andrew argues that product management becomes the bottleneck instead of engineering (however IMO it's more nuanced as we still see good old PRs being the bottleneck still...). Quite interesting to see these very bold and optimistic takes but what is most important is to ensure we can actually quantify the productivity gains vs qualitative metrics. |
|
|
|---|
|
GetDX: Measuring AI Productivity Cracking how to measure AI productivity is something most orgs are trying to figure out, which is why it's great to see this new full-stack framework from DX on measuring AI productivity: The DX AI Measurement Framework gives leaders what feels like a refreshing perspective on AI productivity, as it blends traditional software metrics with recent metrics used for AI copilot productivity; namely this is a three dimension scorecard looking at utilization, impact, and cost. This basically dives into the existing productivity metrics, and seeing improvements from other tech giants, such as Booking.com lifting throughput 16%, Intercom seeing 41% jump in AI-saved developer hours, as well as many others. |
|
|
|---|
|
Coding Languages by Efficiency In production ML, the language wrapping your models can swing your power bill and latency SLOs by an order of magnitude - this study provides an interesting breakdown of efficiency per language: This is quite an insightful study that benchmarks 27 mainstream programming languages on ten CPU-bound Computer Language Benchmarks Game tasks with Intel RAPL showing various pretty insightful results. Unsurprisingly, compiled languages average 120j and 5s per task, while VM-based and interpreted languages burn 576 J and take 20-99s (~20× energy & ~17× latency spread). As expected C, Rust and C++ lead the way with ~57–77j, 2–3 s; followed by Ada (?) and Java; and indeed followed at the bottom by languages like Perl (still alive!), Python, Ruby, JRuby and Lua (2,660–4,600J, 94–167 s). |
|
|---|
|
Twitter Releasing Grok Model It has been surprisingly to see Twitter / X releasing Grok recently with benchmarks that seem surpass all other foundation models out there; however it takes more than raw power to win the race: Grok4 packs a 10× jump in pre-training compute with another 10× in RL-on-reasoning tasks, achieving pretty impressive results with SoTA results on the top benchmarks for LLMs. They are doubling down with features as well such as tool call integration now natively trained into the policy for more dependable chain-of-thought executions. It is interesting however to see that as of these days most of these services are now converging in both capabilities but also functionality, so it will take more than just getting the best benchmarks to beat the competition. |
|
|
|---|
|
Decreased Productivity with AI We see surveys showing improvements in productivity with AI, but this survey actually shows the opposite; decreased productivity with AI copilots: A field experiment with 16 experienced contributors was carried out on large OSS repositories (~1m lines of code). This encompassed 246 issues with labels as "AI-allowed" or "AI-disallowed" for fair comparison, and the outcomes showed that early-2025 coding assistants (Cursor Pro + Claude 3.5/3.7) slowed developers by 19% even though both participants and experts had forecast substantial speed-ups. The drag appears to stem from unreliable suggestions, context-window limits on sprawling codebases, and tacit quality conventions that demand extra review and rework. This is an experience that tends to resonate with a lot of us; the potential still remains however there is further nuance on productivity gains beyond superficial speed on lines of code. |
|
|
|---|
|
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
