Welcome to Nural's newsletter where we explore how AI is being used to tackle global grand challenges.
As always in the newsletter you will find a compilation of articles, news and cool companies all focusing on using AI to tackle global grand challenges.
- A piece I wrote on the Montreal AI Ethics Institute's site which explores AI and healthcare
- Researchers have found significant flaws in research being produced which attempts to use AI to diagnose COVID-19
- Professors are turning down Free money from Google due to their previous treatment of the Ethics team lead, Timnit Gebru
- How AI can unlock renewable energy grid resilience
- and more...
If you would like to support our continued work from £1 then click here!
Nural Article - Artificial Intelligence and Healthcare: From Sci-Fi to Reality
(This article was written as a guest article for the Montreal AI Ethics Institute)
Key Recent Developments
Major flaws found in machine learning for COVID-19 diagnosis
What: Researchers from Cambridge and Manchester Universities have assessed over 2000 papers produced which attempt to use ML to diagnose COVID-19 or help with therapy. They have found serious shortcomings which highlight the existing weaknesses of using ML in healthcare. Most papers made no attempt to "perform external validation of training data, did not assess model sensitivity or robustness, and did not report the demographics of people represented in training data."
Key Takeaway: The transition from AI research to AI in a clinical setting requires better quality data which is evaluated in a more rigorous way than is currently done in most papers. Bridging this implementation gap is often expensive or requires a change in focus from searching for the best AI algorithms possible to searching for the most effective algorithm for mass adoption.
Stop Calling Everything AI, Machine-Learning Pioneer Says
What: Michael Jordan (not the basketball player), who is a researcher at the University of California, Berkeley has shone a light on the widespread misunderstanding of what AI is and its limitations. He believes that AI works best to augment human intelligence in specific tasks and that people should refrain from talking about AI as if there is some kind of intelligent thought in computers.
Key Takeaway: Increasingly there has been a blurring of what the term AI truly encapsulates, with many jokes often being made that what was once just statistical analysis is now called machine learning and AI. However, it's important to note that while AI is achieving phenomenal results in certain specific tasks, it would be a mischaracterisation to believe that machines think in the same way humans do. There are countless examples of the systems failing or experiencing huge drops in performance when in new, unexplored contexts.
Why artificial intelligence is key to renewable energy grid resilience
- The global transition to renewable energy will need artificial intelligence (AI) technology to manage decentralized grids.
- AI can balance electricity supply and demand needs in real-time, optimize energy use and storage to reduce rates.
- Technology governance will be needed to democratize access, encourage innovation and ensure resilient electricity sources.
🚀 A 5-step guide to scale responsible AI (World Economic Forum)
Other interesting reads
Cool companies I have come across this week
Text IQ - AI-Powered Privacy, Security, and Data Discovery.
Identify and protect sensitive information hidden in enterprise data
The Language Interpretability Tool (LIT) - The Language Interpretability Tool (LIT) is for researchers and practitioners looking to understand NLP model behavior through a visual, interactive, and extensible tool.
AI/ML must knows
AutoML - The process of automating the process of applying machine learning to real-world problems. AutoML covers the complete pipeline from the raw dataset to the deployable machine learning model.
Transfer Learning - Reusing parts or all of a model desinged for one task on a new task with the aim of reducing training time and improving performance.
Tensorflow/keras/pytorch - Widely used machine learning frameworks
Generative adversarial network - Generative models that create new data instances that resemble your training data. They can be used to generate fake images.
Deep Learning - Deep learning is a form of machine learning based on artificial neural networks.
Nural Research Founder
If this has been interesting, share it with a friend who will find it equally valuable. If you are not already a subscriber, then subscribe here.
If you are enjoying this content and would like to support the work financially then you can amend your plan here from £1/month!