🤩Musk’s surprise move!

Reading time: 5 minutes  

 Newsletter | News | Tools | Grow My Tool  

Today we will discuss-

• 🧠China to test neurosurgery AI

• 🥳Musk’s xAI to open-source Grok this week

• 😳LLMs dangerously biased against African Americans, study finds

• ⚙️9 amazing AI tools you might not have heard of

All this and more - Let's dive in!

👩‍🍳What’s cooking in the newsroom?

China to begin testing custom AI assistant for neurosurgeons

🏥News - In the coming months, China will be testing an AI assistant specifically tailored for neurosurgeons at seven hospitals located in Beijing and other cities. This initiative is part of the government's broader efforts to harness the potential of AI technology in various sectors.

👨🏻‍🔬How was the AI developed? The AI assistant, created by a Hong Kong-based agency affiliated with the Chinese Academy of Sciences, draws its foundation from Meta's Llama 2.0 model.

According to reports, researchers trained and refined the model using papers, medical journals, and manuals, enabling it to serve as a sort of surgical consultant for doctors.

👨🏻‍⚕️How exactly will the AI 'assist' doctors? The AI assistant, named the CARES Copilot 1.0, is designed to respond to questions with references from a vast database of over a million academic records. Researchers claim it will also be able to analyze diagnostic data from MRI, ultrasound, or CT scans, as well as process various forms of media such as images, text, and audio.

In the long run, the researchers envision the AI playing a more proactive role, such as alerting doctors about the potential risks associated with certain procedures.

Elon Musk says his AI startup, xAI, will open-source its ChatGPT rival, Grok, this week

🤗News - Elon Musk has announced that his AI startup, xAI, will open-source its ChatGPT rival, "Grok," later this week. This announcement follows Musk's recent legal action against OpenAI, accusing the organization of shifting away from its original non-profit mission in favor of a for-profit model.

Although Musk didn't provide details on which aspects of Grok he intended to open-source, this decision could enable the public to freely experiment with the technology's code. It also positions xAI alongside companies like Meta and Mistral, both of which have released their AI models as open-source projects.

📢The open-source debate - Musk has been a vocal advocate for open-source projects for years now. And in yet another jab at the ChatGPT maker, Musk said on Monday, "OpenAI is a lie," reiterating his criticism of the Altman-led firm.

However, Musk isn't alone; his lawsuit against OpenAI has ignited a debate among many technologists and investors about the merits of open-source AI.

The long and short of it is, while making technology open-source can drive innovation forward, experts are concerned that open-source AI models might be exploited by terrorists for creating chemical weapons or even to develop a conscious super-intelligence beyond human control.

Cornell study finds LLMs still discriminate against African American English speakers

🙇🏼‍♂️News - A new study by Cornell University suggests that large language models (LLMs) are prone to showing bias against speakers of African American English. 

The research indicates that the dialect individuals speak can affect how AI algorithms perceive them, leading to biased judgments regarding their characteristics, employability, and potential involvement in criminal activities.

🧐What was the basis of the research? The study examined language models like OpenAI’s ChatGPT and GPT-4, Meta’s LLaMA2, and French Mistral 7B. Through a method called "matched guise probing," researchers presented prompts in both African American English and Standardized American English to these models. They then studied how the models perceived various characteristics of individuals based on the language used in the prompts.

🕵️‍♂️What did they find? 

• According to the findings, GPT-4 technology is more likely to recommend death sentences for defendants who use English associated with African Americans, even when their race is not mentioned.

• The study also discovered that large language models (LLMs) tend to assume that speakers of African American English have lower-status jobs than those who speak Standardized English, even when the speakers' racial backgrounds are not disclosed.

• The study also found that as the size of large language models (LLMs) increases, their understanding of African American English improves, and they become less likely to use explicitly racist language. However, the size of the LLM does not alter its underlying biases. What this simply means is, LLMs have learned to "superficially conceal" their underlying racial biases while maintaining them at a deeper level.

🙆🏻‍♀️What else is happening?

• AI expert says Kate Middleton's 'comeback' photo likely to have been 'tweaked’

• Dozens of top scientists sign effort to prevent A.I. bioweapons

• Nvidia sued by authors over AI use of copyrighted works

• Microsoft begins blocking some terms that caused its AI tool to create violent, sexual images

• OpenAI announces new board members, reinstates CEO Sam Altman

• AI talent war heats up in Europe

👨‍🎓Interesting facts

• UCLA bioengineers created an AI-powered ASL glove that could translate 660 sign gestures to speech via a smartphone app, enhancing communication between signers and non-signers.

• The COMPAS algorithm, once used for predicting criminal behavior in sentencing, faced accusations of racial bias in 2016. This sparked discussions on AI ethics, emphasizing the need for unbiased algorithms and transparency.

• In 2017, the faculty at MIT created an AI-enabled Twitter horror story writer named Shelly. She drew inspiration from 140,000 stories on Reddit to craft spooky story prompts. Users on Twitter used to complete these via replies on the same thread.

👩🏼‍🚒Discover mind-blowing AI tools

1. Osum -A tool that provides instant access to detailed research, SWOT analysis, buyer personas, growth opportunities, and more for any product or business (Free trial)**

2. QR Galaxy - A platform that allows users to create unique QR codes inspired by images of their choice ($4/per QR code)**

3. SkinGenerator.io - A website that allows users to create custom skins for video game characters ($0.99/5 skins+ 50 additional skins at $0.02/ generation)

4. Artguru - An AI-powered tool that allows users to create highly realistic and expressive avatars and portraits in various styles ($2.99/20 high-res avatars from choice of 6 styles)

5. Salesforge - A comprehensive sales execution app that streamlines email deliverability and personalized cold email outreach ($48/month)

6. Architect Render  - An AI-powered platform that allows users to transform photos, sketches, and designs into realistic 3D renderings ($5/5 credits) 

7. Rizemail - An AI-powered email summarization tool that helps users get to the core of their unread newsletters and long email threads (Free)

8. Humata - A tool that provides an easy way to quickly summarize long papers and get instant answers ($1.99/month)

9. Remini -  A photo-enhancement app that uses AI to enhance everyday portraits and selfies to achieve a professional and influencer-style output ($17.99/month)

🌟Mythbuster

Myth - AI art is limited to copying styles

Fact - AI's creative potential goes beyond imitation. It replicates styles but can also generate fresh aesthetics. Advanced techniques like GANs blend styles, spark inspiration and push artistic boundaries. AI amplifies creativity, allowing art that would be impossible to express through conventional means.

[The tool marked with ** is a sponsored feature.]