AI Bits
Posts
💻Ai competition is fierce: Elon, Google, OpenAI and Anthropic

💻Ai competition is fierce: Elon, Google, OpenAI and Anthropic

Catch up with this week's AI news and learn about another AI tool.

A.I. Bits
July 18, 2023

Hey everyone!

Virtual Visions 1 results will be out in a few days so stay tuned to see the winners! It just could be you! 👀

Here’s what you’ll learn about today:

Musk launches his new AI company, xAI 🚀
Claude 2 vs Google Bard ⚔️
Beijing publishes its AI governance rules 🌆
A detailed and thorough AI researcher: Aomni🖼️

Reading time: 4 min 39 seconds

AI artists this is for you 👇

➡️Check out AI BIts DIscord ⬅️

Join a buzzing community of AI artists to learn and create together!

News Reel:

Highlight 1: Elon Musk has launched his new AI company - xAI!

Elon working on his new project

Elon Musk has launched a new AI startup called xAI, staffed by top researchers, with the ambitious goal of developing a "theory of everything" to advance neural networks and understand the true nature of the universe. 🤯 The bits:

More details are to be shared in Twitter Spaces chat on Friday. 🗓️
xAI team includes alumni from DeepMind, OpenAI, Google, Microsoft, and Twitter.
xAI openly competes with OpenAI, Google, and Anthropic.
xAI secured GPUs from Nvidia to power potential large language models. 🖥️
The advisor, Dan Hendrycks, is from the Center for AI Safety focused on AI existential risk.
In the future, xAi might work with Twitter, and Tesla.

Find their website here.

Highlight 2: New Updates for Google Bard and its competitor: Anthropic’s Claude 2!

Anthropic has released Claude 2, an upgraded conversational AI assistant focused on safety, reasoning abilities, and summarization of lengthy texts; while Google has expanded the availability of its AI chatbot Bard to over 40 languages while adding new features like image processing and conversation sharing capabilities. 💻Here are the bits:

Anthropic - Founder of Claude

Claude 2:

It is designed for helpful conversations and has improved conversational abilities.
Claude 2 excels at coding, math, and reasoning compared to their predecessor
Allows input/output of up to 100,000 tokens, great for long documents.
Fine-tuned to reduce harmful outputs - twice as effective as Claude 1.3.
Uses "Constitutional AI" based on principles like the human rights declaration. 🧾
Can summarise up to 75,000-word texts but prone to factual errors.

Google Bard

Bard Updates:

Available in 40+ new languages including Arabic, Chinese, German, Hindi, Spanish
Expanded access to 27 EU countries and Brazil.
Can now upload images with Lens to boost creativity
Added text-to-speech in 40+ languages including Hindi, Spanish
Can pin and rename conversations for continuity
Shareable links allow others to view conversations
New options to modify responses to be simpler, longer, etc.
Can export Python code to Replit in addition to Google Colab
Great at HTML and CSS, takes in a website image and returns the HTML and CSS to build it 🤯

In summary, Bard has more languages, and creative features like images/voice, conversation management, and response modification. Claude 2 leads in transparency around safety and reasoning abilities.

Read the article for Bard updates here and for Claude 2 here.

Try Claude 2 here! (Only available in the US and the UK - don’t worry, just use a VPN)

Highlight 3: Beijing publishes its AI governance rules!

The Future of Beijing - 2100

China has unveiled new rules governing generative AI that aim to align services with socialist values while also expanding public data resources, though they impose stricter content restrictions than few other countries. The bits:

Operators must ensure services adhere to socialism and avoid content that undermines state power or national unity. 🚫
Services are prohibited from promoting harmful content like ethnic hatred, violence, and obscenity.
China aims to develop public training data platforms and share model-making hardware.
AI should use secure, proven tools like chips, software, and computing power.
Consent is needed for personal data, focus on data quality and diversity.
Algorithms must not discriminate based on ethnicity, gender, etc. 💙
Most operators need licenses, adding regulatory oversight.
Rules take effect August 15, 2023, and will benchmark global AI governance discussions.

You can find the full copy of the regulations here (It’s in Chinese though). Read more here.

A.I. Showcase: Aomni ai

Aomni ai Logo

I haven’t done any research from scratch for uni since I learned about Aomni. Even if it doesn’t provide a whole research report, it’s an amazing place to start your research and all you need to do is write your research question. Here’s how 👇🏽:

1. Go to Aomni and click on the ‘Query box’

2. Write your Research Question

3. Click on Submit - that's it! Feel free to leave the page as Aomni will send you an email in a few seconds once it has written your summarised research report for you!