US lawmakers have proposed the Algorithmic Accountability Act, which, if passed, would ask the Federal Trade Commission to assess whether AI algorithms and their training data are biased or discriminatory, and whether they pose privacy or security risk to consumers. This is in the same spirit as the EU guidelines on AI we covered last week, but more focused and concrete. To address the potential anti-competitiveness complying with such regulations may introduce, the bill would only apply to “companies that make over $50 million a year, hold information on at least 1 million people, or primarily act as data brokers that buy and sell data.”
Universities in the UK like Staffordshire and Bolton College have deployed AI chatbots in recent years that can help students answer questions such as ones about class schedules and homework deadlines. While current AI chatbots are highly limited in their use cases - they have to be preprogrammed to answer specific types of questions - many educators are optimistic about AI for education, citing the potential for AI to supplement teachers and “reduce their administrative workload so they can focus on more creative or theoretical aspect of their courses.”
Concerns over a new AI “arms race” among the world’s most powerful countries has been brewing for some time now, but this may distract us from a very real danger with AI development: rushing it out before it’s safe and truly ready. It may be tempting to demonstate and deploy new technology fast to remain on the cutting edge, but past experience with computers has shown that can lead to security flaws and bugs, which is even more likely with AI technologies given how difficult they are to fully understand and test. So rather than upping the stakes and pace prematurely, we must acknowledge the challenges of developing and deploying AI technology and make sure to invest in it being done safely and strategically.
“But the emerging narrative of an “AI arms race” reflects a mistaken view of the risks from AI—and introduces significant new risks as a result. For each country, the real danger is not that it will fall behind its competitors in AI but that the perception of a race will prompt everyone to rush to deploy unsafe AI systems. In their desire to win, countries risk endangering themselves just as much as their opponents.”
Advances & Business
- Microsoft wants to unleash its AI expertise on climate change — “In a broad new set of sustainability commitments, the company wants to use its tech to develop tools to monitor and find insights in environmental data.”
- Artificial intelligence is helping old video games look like new — “Modders are taking advantage of AI tools to update old graphics”
- Folding Revolution — New deep-learning approach predicts protein structure from amino acid sequence
- Boston Dynamics debuts the production version of SpotMini
Concerns & Hype
- ‘Disastrous’ lack of diversity in AI industry perpetuates bias, study finds — “Report says an overwhelmingly white and male field has reached ‘a moment of reckoning’ over discriminatory systems”
- We’ve been warned about AI and music for over 50 years, but no one’s prepared — “AI is capable of making music, but does that make AI an artist? As AI begins to reshape how music is made, our legal systems are going to be confronted with some messy questions regarding authorship.”
- When ML and Data Science are the death of a good company: A cautionary tale.
- IBM halting sales of Watson AI tool for drug discovery amid sluggish growth
Analysis & Policy
- Some AI just shouldn’t exist — “Attempts to “fix” biased AI can actually harm black, gay, and transgender people.”
- Standards for AI Governance: International Standards to Enable Global Coordination in AI Research & Development
- How will AI change your life? AI Now Institute founders Kate Crawford and Meredith Whittaker explain.
- How to Train Your OpenAI Five
- Uncertainty Autoencoders: Learning Compressed Representations via Variational Information Maximization
- Representer Point Selection for Explaining Deep Neural Networks
- Gender, Race, and Power in AI
That’s all for this week! If you are not subscribed and liked this, feel free to subscribe below!