• Not A Bot
  • Posts
  • 🌊 Google’s AI can predict floods

🌊 Google’s AI can predict floods

PLUS: The newest, biggest dataset to date

New here? Welcome to NAB, and make sure to subscribe for more!

Happy Tuesday, fellow humans. 👋

In today’s edition of Trivia Tuesday: What do you call a group of unicorns? 🦄 

The answer will be at the end of the newsletter :)

Anyway, let’s dive into it…

Have anything you’d like to share with over 40k subscribers? 

🧵 In today's edition:

  • 🌊 Google’s AI can predict floods

  • 🧑‍🏫 Meet Khanmigo

  • 😲 The biggest dataset yet

  • 🤑 AI Fundraising News

🤖 Top AI News

🌊 Google’s AI can predict floods

Most of the world lacks river flow measurements. This makes predicting floods difficult, especially in lower-income countries that lack data.

In 2018, Google launched an AI flood forecast system for India and Bangladesh that was excellent in predicting when floods were incoming.

Now, Google's Grey Nearing and colleagues have proven their AI can forecast floods worldwide. They trained it on public river flow data and weather forecasts gathered by the World Meteorological Organization, allowing it to learn general flood relationships to predict future events. Moreover, unlike traditional models, it transfers knowledge between locations.

The results?

The AI was tested on 5,000 global water measurements from 1984-2021. Its 4-day predictions matched the best conventional forecasts on the same day.

Plus, for data-poor South America and Africa, it predicted floods 4-7 days out.

Quite extraordinary.

Overall: Google now covers 80+ countries on its Flood Hub and Search and has sent flood alerts since October 2022. The versatile AI can now predict floods far in advance, even without abundant data, significantly advancing global disaster forecasting.

Read more: New Scientist

🧑‍🏫 Meet Khanmigo

AI tools like ChatGPT are sparking fears of cheating, but banning them from schools might be an overcorrection.

In an attempt to sustainably bring AI to the classroom, Khan Academy is introducing its AI tutor Khanmigo to over 8,000 students year. It offers personalized math, science, and writing help, as well as debates and chats with historical figures.

Khan Academy trained GPT4 on its own content to create Khanmigo, but its functionality is slightly different - the system won't give students direct answers. Instead, it allows real-time user feedback to help fix mistakes.

It also benefits teachers, allowing them to create lessons and identify struggling students while providing them with chat history.

Khanmigo isn't immune to AI "hallucinations." It struggles with math calculations, sometimes congratulating the wrong answers. But it shines at walking students through problems with hints and questions.

Overall: Foolproof prevention of AI cheating is impossible, but creative, critical assignments can minimize abuse.

Rather than banning classroom AI, teachers must equip students to use it wisely. The technology's rise seems inevitable, and with thoughtful integration, AI has the potential to engage students in new, rewarding ways.

Read more: CNN

😲 The biggest dataset yet

Powerful language models like GPT-4 are shrouded in secrecy. Companies like OpenAI publish basic stats about their datasets, but the sources and processes behind them remain proprietary, often fueling speculation about unethical data sourcing.

To challenge the status quo, the Allen Institute for AI (AI2) is introducing Dolma - a massive new free and transparent dataset.

Dolma will train AI2's upcoming open language model, OLMo, allowing all the underlying data to be modified by researchers. All 3 billion tokens worth of sources and processing are fully documented, enabling research replication and oversight.

Of course, no dataset will be 100% pristine. But AI2 strives for ethical sourcing and maximum openness. This addresses public concerns about Big Tech’s lack of transparency re: language models and enables computer scientists worldwide to inspect and build on Dolma freely.

Overall: AI advances fastest through collaboration. With privacy guarded and sources transparent, Dolma brokers trust in a field where technology races ahead of social consensus.

Its unprecedented openness sets a new standard for integrity in language data, and it will be interesting to see if big tech adopts this mindset.

As usual, you can access Dolma through HuggingFace.

Read more: TechCrunch


Taplio: Build your personal brand with AI 🚀

Imagine this:

You are trying to level up your LinkedIn game, but you don’t have the time to develop eye-catching, engaging posts that will allow you to build your brand.

If that feels familiar, you must check out Taplio, the 🤖 AI-powered tool trusted by 3000+ professionals.

  • 💡 Create high-quality posts 10x faster

  • 🗓️ Schedule and analyze content

  • 👥 Build relationships

All in one place!

Get inspiration from viral posts, an AI writing coach, and bulk messaging to grow your brand effortlessly.

💰 Plus, there's zero risk with a free 7-day trial and a 30-day money-back guarantee!

Isn't it time you made the most of the 850M+ LinkedIn users?

Try it free and become the LinkedIn influencer you were meant to be!

🤑 AI Fundraising News

Genesis raises $200 million in Series B funding for AI drug discovery research.

Writerly AI Raises $2M in initial funding to accelerate the development and expansion of EKOM, a platform that generates and optimizes product detail page assets for e-commerce stores.

🗞️ AI Quick-Bytes

What else is going on?

  1. Tricks for making AI chatbots break the rules are freely available online

  2. Singapore workers are the world’s fastest in adopting AI skills

  3. Korea’s Internet Leader to Unfurl Entry in ChatGPT-Style AI Race

  4. YouTube wants to benefit from AI-generated music without the copyright headaches

  5. How any SaaS company can monetize generative AI

  6. Arm IPO to put SoftBank's AI hard sell to the test

  7. A message to OpenAI's Sam Altman was an AI startup founder's last-ditch effort to help save her company — and it worked

  8. Baidu’s Results May Shake Optimism Over China AI: Tech Watch

🐦️ Tweet of the day

🧑‍💻 Coding…made 100x easier

What did you think of today's newsletter?

Login or Subscribe to participate in polls.

Trivia Tuesday answer: A group of unicorns is called a “Blessing” 🙏 😇 

And that does it for today's issue.

As always, thanks for reading. Have a great day, and see you next time! ✌️

— Haroon: (definitely) Not A Robot and @haroonchoudery on Twitter

P.S.: Check us out on our socials - snacks on us if you also hit follow :)

Join the conversation

or to participate.