Welcome to Nural's newsletter focusing on how AI is being used to tackle global grand challenges.
Packed inside we have
- Introducing The World's Largest Open Multilingual Language Model: BLOOM
- England’s health service will use drones to deliver vital chemotherapy drugs
- and Meta’s ‘Make-A-Scene’ Tech Is Pushing the Boundaries of AI-Generated art
If you would like to support our continued work from £1 then click here!
Marcel Hedman
Key Recent Developments
Introducing The World's Largest Open Multilingual Language Model: BLOOM
What: BigScience initiative have released BLOOM, the first multilingual LLM trained in complete transparency (claimed to be as powerful as GPT-3) — the result of the largest collaboration of AI researchers ever involved in a single research project.
With its 176 billion parameters, BLOOM is able to generate text in 46 natural languages and 13 programming languages. For almost all of them, such as Spanish, French and Arabic, BLOOM will be the first language model with over 100B parameters ever created.
The model is available in open source along with the code and datasets used to create it.
Key Takeaway: The release of the model represents a pivotal moment across multiple dimensions. Firstly, it is a step towards democratizing top-performance AI away from the large tech firms who are typically the only ones with the resources to build at this scale. Secondly, it is a step away from the siloing of top performance AI solely on the English language. A multilingual model of this scale opens up a realm of possibilities in multiple regions which have not had access to a local dialect model previously.
A lot of potential Geo-arbitrage opportunities...
England’s health service will use drones to deliver vital chemotherapy drugs
What: The UK NHS (national health service), has announced that it will test delivering vital chemotherapy drugs via drone to the Isle of Wight. The body has partnered with Apian, a drone technology startup founded by former NHS doctors and former Google employees. Test flights are due to begin shortly, and it’s hoped that the system will reduce journey times for the drugs, cut costs and enable cancer patients to receive treatment far more locally.
Key Takeaway: Chemotherapy medicines have a short shelf life, meaning that patients often have to take long journeys to hospitals to receive the medication as lengthy home deliveries would not work. Drones unlock a new realm of cost and time savings for patients and health providers. In this case, a 4h journey is cut to only 30 minutes.
It's encouraging to see nationally funded health services pushing innovation in this way.
Meta’s ‘Make-A-Scene’ Tech Is Pushing the Boundaries of AI-Generated art
Francesca Aton
What: Meta AI published a report last week on its artificial intelligence (AI) text-to-image generation research. The technology, dubbed Make-A-Scene, allows users to draw a freeform digital sketch to accompany a text prompt. The AI then uses the two together to produce an AI-generated image.
This is in contrast the the recent Dall-E mini AI image generator which only takes in text as an input.
Key Takeaway: It's incredible to see the model operate within the constraints of both the text and the the offered sketch. It'd be interesting to see what would happen if the sketch somehow contradicted the input text.
These developments in AI generated art are important because they will fundamentally affect how we perceive creativity, and the validity of information we see. The examples below are generating cartoon outputs, but applying this to real images unlocks a whole realm of new avenues... some good and some less so.
AI Ethics
🚀 $71,000 to find flaws in publicly deployed or released AI systems
🚀 Genetic Screening Now Lets Parents Pick the Healthiest Embryos
🚀 Self driving cars frozen in traffic and trapping human drivers
🚀 Why business is booming for military AI startups - The war in Ukraine
Other interesting reads
🚀 How Spotify Uses Semantic Search for Podcasts
🚀 Artificial intelligence model finds potential drug molecules a thousand times faster
🚀 Deep Learning with a Small Training Batch (or Lack Thereof)
🚀 Job Hunt as a PhD in AI / ML / RL: How it Actually Happens
Cool companies found this week
Health
Wysa - Helping users self-manage stressors by blending AI-guided listening with professional expert support. They have recently raised $20m in series B funding.
Climate
Monolith - Monolith produces carbon-free hydrogen that provides for the high-energy needs of the growing world — but at the affordable cost the world needs to make it work. (This is not directly AI related but interesting to see commercial carbon-free hydrogen power taking off)
AI/ML must knows
Foundation Models - any model trained on broad data at scale that can be fine-tuned to a wide range of downstream tasks. Examples include BERT and GPT-3. (See also Transfer Learning)
Few shot learning - Supervised learning using only a small dataset to master the task.
Transfer Learning - Reusing parts or all of a model designed for one task on a new task with the aim of reducing training time and improving performance.
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.
Best,
Marcel Hedman
Nural Research Founder
www.nural.cc
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!