🪄 Meta + Microsoft = Magic

PLUS: Reports warn that UK’s approach to AI safety lacks credibility

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

Happy Wednesday, fellow humans. 👋

Candidly speaking, the opposite of Artificial Intelligence is Human Intelligence.

Literally speaking, the opposite of Artificial Intelligence is Real Stupidity.

So if we are to candidly speak literally…Human Intelligence = Real Stupidity??

Sorry… not sure where I was going with this.

Let’s just dive into it…

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

🧵 In today's edition:

  • 🪄 Meta + Microsoft = Magic

  • 💰️ Infosys signs $2B AI deal

  • 🥊 France vs Britain for Europe’s A.I. crown

  • 🤑 AI Fundraising News

🤖 Top AI News

🪄 Meta + Microsoft = Magic

Yesterday was a BIG day for LLMs and open-source advocates.

Meta released Llama-2, a new LLM model that, so far, seems to perform almost as well as GPT 3.5 and outperforms other open-source models on various external benchmarks, including reasoning, coding, proficiency, and knowledge tests.

Here are some key deets:

  • Llama 2 was pre-trained on 2 trillion tokens, almost twice more than Llama. Most of the data was pulled from publicly available sources.

  • Llama 2 comes in two types - Llama 2 and Llama 2 Chat

  • Llama 2 chat was fine-tuned for 2-way conversations

  • Llama 2 and Llama 2 Chat come in 3 different model sizes - 7B, 13B, and 70B.

  • Although it is still significantly behind on coding capabilities compared to GPT4 and PaLM2, Llama 2 answered on par across a set of 4,000 prompts testing for “helpfulness” and “safety.”

  • Almost half of their white paper is dedicated to AI safety issues, discussing safety guardrails and evaluations. 👏 

And in an interesting turn of events, Microsoft announced the support of Llama 2 on Azure and Windows at Microsoft Inspire, which enables developers and organizations to build generative AI-powered tools and experiences.

MS might not be as exclusive with OpenAI as we once thought.

Azure customers can now fine-tune and deploy the different-parameter Llama 2 models on Azure, and Llama will be optimized to run locally on Windows.

Overall: Open source might eventually prove to be better than closed-source (TBD), but the way Meta has approached this release might be a winning strategy.

They launched it with multiple partners and a written Statement of Support co-signed by multiple heavy hitters in the industry.

Is OpenAI in trouble?

Read more: TechCrunch | Meta Blog

💰️ India's Infosys signs $2B AI deal

India’s 2nd largest software provider, Infosys, signed a deal with an “unnamed existing client” to provide AI and automation services.

No idea why they are being so secretive, but sure - we’ll roll with it.

The agreement will last five years and is estimated to have a target spend of $2bn.

This could be a continuation of Infosys Topaz, their generative AI platform launched earlier this year.

The deal includes development, modernization, and maintenance services related to AI and automation and comes as businesses globally invest more heavily in such technologies.

Read more: Economic Times

🥊 France vs Britain for Europe’s A.I. crown

France and the UK are in a game of chicken - but instead of seeing who’s scared, this is to see who can throw the most amount of $$$ at AI development.

Both countries are competing to be Europe's AI capital, with each country making bold statements about their potential to lead the market.

French President Emmanuel Macron plans to invest $560M (€500 million) in AI "champions," adding to previous government commitments to pump €1.5 billion into the technology by 2022.

Meanwhile, the UK government has pledged $1.3B (£1 billion) to supercomputing and AI research to build an "exascale" computer that would rival OpenAI's generative AI chatbot.

Overall: While all this money is being poured into AI, policymakers in the UK and France are taking different regulatory approaches, which could impact the countries' overall attractiveness for AI investment.

What do you think happens?

Who comes out on top?

Login or Subscribe to participate in polls.

Read more: CNBC

🤑 AI Fundraising News

Cybersecurity firm Netcraft lands $100M investment in its first funding round since its founding in 1995.

Tractable snaps up $65M led by SoftBank for car and property damage appraisals using AI

🗞️ AI Quick-Bytes

What else is going on?

  1. Most outsourced coders in India will be gone in 2 years due to A.I., Stability AI boss predicts

  2. SAP Advances Vision of Business AI with Investments in Aleph Alpha, Anthropic, and Cohere to Complement $1+ Billion AI Commitment from Sapphire Ventures

  3. Microsoft will charge businesses $30 per user for its 365 AI Copilot

  4. Ron DeSantis Super PAC Unleashes AI Attack Ad Against Trump

  5. DataStax brings vector search to its Astra DB database service

  6. China AI Chip Firm Targeting Nvidia Seeks HK IPO in 2023

  7. UK’s approach to AI safety lacks credibility, report warns

  8. Hollywood’s writers and actors are on strike

🐦️ Tweet of the day

💭 Pretty interesting thoughts in the responses to this tweet

What did you think of today's newsletter?

Login or Subscribe to participate in polls.

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.