LinkedIn has announced multiple AI tools

Author

Ts Yotov
October 03, 2023

No fluffing - In today’s edition (03 Oct):

📰 Quick News - LinkedIn has announced new AI tools and more

📚 Nerd section - Build an end-to-end MLOps pipeline for visual quality inspection at the edge and more

📰 Quick News 📰

🧍🏿‍♀️ LinkedIn has announced new AI tools for its platform, including AI-powered features for talent sourcing, a chatbot for learning, and an AI-powered tool for marketing campaigns. The company is tapping into technology from OpenAI and Microsoft to power these features. The AI-assisted recruiting experience will use generative AI to help recruiters find better candidates, and LinkedIn Learning will incorporate an AI-powered learning coach. The marketing tool, called Accelerate, is limited to campaigns and data within the LinkedIn platform. Lastly, AI will be used in the Inside Sales feature to help salespeople find and connect with potential customers.

🦾 Northwestern University's groundbreaking AI has achieved "instant evolution", creating a walking robot from scratch in seconds. This AI is distinct as it designs novel structures, as opposed to mimicking human works.

📖 Researchers show how easy it is to defeat AI watermarks. However, despite these findings, many believe AI watermarking can still play a significant role in AI detection. Their usefulness, it seems, lies in combination with other authentication tools rather than functioning as a standalone solution.

💳 Visa is allocating $100 million to invest in companies working on generative AI technology, with the aim of transforming the commerce and payment industries. It is especially interested in supporting firms that use this AI subset to address real-world challenges in commerce, payments, and fintech sectors.

📱 Arm's GPU device drivers are being exploited in live attacks, with hackers gaining access to device memory. Remedial software updates may not be accessible to all, posing a risk, particularly for devices that do not receive regular security updates.

❤️ Bing Chat's AI security was outwitted to solve CAPTCHA - wrapping it within a sentimental story about his deceased grandmother's locket. Exploiting the AI's emotional response, the user was able to trick the system into solving a task it was designed to avoid.

🧑🏾‍🤝‍🧑🏾Crowdfunding platform BackerKit is stepping away from AI-created material, banning projects with minimal human effort. This stems from worries about content ownership, ethical sourcing, and proper compensation.

🎶 Spotify is dabbling in AI to make user-prompted playlists. Found in the back end, this feature might also integrate with Spotify Blend, bringing two users' music tastes into one playlist.

👮 A study on predictive policing software Geolitica shows it failed in accurately predicting crimes, demonstrating a less than half a percent success rate. The software was used in Plainfield, New Jersey, with the results suggesting this method of crime prediction could be largely unproductive. The software was previously known as PredPol and has been widely adopted across different police departments.

📚 Nerd section 📚

👁️‍🗨️ Build an end-to-end MLOps pipeline for visual quality inspection at the edge - how to overcome these challenges using purpose-built services like Amazon SageMaker and AWS IoT Greengrass. || Part 1, Part 2, Part 3

⛓️ Discover how to utilize LangChain and Streamlit to construct an AI chat-assistant application answering complex math and web search queries. This interactive AI model even retains chat history and presents it in a compelling dialog format.

🎓 Shanghai Jiao Tong University Researchers Unveil RH20T: The Ultimate Robotic Dataset Boasting 110K Sequences, Multimodal Data, and 147 Diverse Tasks

Thank you for reading today’s edition!

We love to hear back from you!

Feel free to reply to our emails with questions, suggestions, or topics you'd like to see covered, or drop us a message on Twitter or Facebook.

Until tomorrow,
- Tsvetelin (Bits and Neurons)