3 Things I Learned About AI This Week

3 Things I Learned About AI This Week

1. AI's Hefty Environmental Footprint

The environmental cost of AI is staggering. Training a single large AI model can generate carbon dioxide emissions equivalent to ~300,000 kg, according to MIT Technology Review. This energy-intensive process contributes significantly to resource consumption, and specifically the burning of fossil fuels.

But that's not all. AI's thirst for resources extends beyond energy. Hyperscale data centers, crucial for AI and cloud computing, use water at an alarming rate. Dgtl Infra reports that these centers can use up to 550,000 gallons of water daily - roughly equivalent to the water consumption of over 1,800 average U.S. families.

2. AI Experts' Concerning Predictions

While I was vaguely aware of potential AI risks, the scale of concern among experts is eye-opening. Ethan Mollick's book Co-Intelligence reveals a stark contrast in expert opinions:

  • AI experts estimate a 12% chance of AI causing a catastrophic event that could wipe out at least 10% of Earth's population by 2100.
  • In contrast, futurists place this risk at a more conservative 2%.

This wide discrepancy underscores the uncertainty and debate surrounding AI's long-term impact on humanity. It's clear that even among those at the forefront of AI development and research, there's significant disagreement about the technology's potential dangers.

3. The Power of Politeness: Complimenting AI for Better Performance

In an unexpected twist, it seems that being nice to AI can actually improve its performance. This insight, which I first heard from a senior AWS strategist, is common in AI circles and on many online forums. The approach is surprisingly simple:

  • Compliment the AI on its intelligence and thoughtfulness.
  • Encourage the AI to think like an expert and follow a step-by-step process.

This positive reinforcement apparently prompts the AI to adopt the persona of an expert, leading to more thorough and considered responses. It's a fascinating glimpse into the psychology of artificial intelligence and how human interaction styles can influence machine performance.

Now, what's one thing you learned this week? (or recently ??)

#learning #ai #artificialintelligence

Betsy Stratman Pruitt

College & Career Counselor, CEO at Education Navigation. Instructor, UC Berkeley Extension College Counseling Certification Program. Expertise in Undergrad Admissions, Graduate School and Allied Health fields.

4 个月

David L. Dimmett, Ed.D. Point number 3 is so interesting... I have told my kids since our first encounters with Siri that it's important to be kind when communicating because that is how Siri will respond...

回复
Patricia Sanderlin, CSPO?, SAFe? POPM

Product Manager @ Project Lead The Way | Education | ?? CPSO? | Certified SAFe? 6 PO/PM ??| Customer Focus | Innovation | AI | Love Horses!

4 个月

I learned recently that generative AI has a long way to go before it would become perfect. In a previous part-time role, I was training AI models (META to be exact) and it was interesting to see how AI struggled with defining emotions or slang. While it excels in combing through data and its algorithms are improving constantly to form conclusions, it had challenges with, for example, sarcasm. To your point of being nice to AI, generative AI learns from you and the more you interact with it, the more closely it will adapt to your style of conversation. So you do yourself a favor being polite to AI! :-)

Jason Rausch

SVP of Instructional Development @ Project Lead The Way | Curriculum Innovation, STEM

4 个月

I have been learning about the potential future impact quantum computing will have on AI. Quantum computing has the potential to revolutionize AI by performing specific computations exponentially faster than classical computers https://techannouncer.com/quantum-ai-how-quantum-computing-will-supercharge-artificial-intelligence/ As we witness such rapid advancements in AI, will we see similar advancements in the next decade around quantum computing?

Martha McCabe

Senior Vice President of Community Partnerships at Project Lead The Way

4 个月

I recently learned that if you want to work in AI, you will need to learn Python. Glad that our Project Lead The Way CS Principals course teaches Python! Preparing students for a STEM-driven world.

要查看或添加评论,请登录

David L. Dimmett, EdD的更多文章

社区洞察

其他会员也浏览了