🤖 The Age of Ultron Begins with Predicting Human Actions 🔮

Read time: 3 mins

In today’s email:

• 🧪 Research highlights: Robots Learn to Predict Human Preferences in Assembly Tasks

• 🚨 Industry news: Meta Unleashes SAM, an AI Model That Can Spot Hidden Figures in Images

• ⚖️ Policy and regulation: OpenAI Takes the High Road in AI Safety with Rigorous Measures

• 🌐 AI and society: Am AI the Drama?: From Optimism to Pessimism

• 🧰 Tool of the day: Customized Responses in Seconds

• ⏰ AI concept of the day: Sense and Semantics: How Jane Austen Would Explain NLP

🔮 Robots Learn to Predict Human Preferences in Assembly Tasks

The University of Southern California (USC) computer science researchers developed a system that teaches robots how to predict human preferences in assembly tasks, a critical skill for robots to become collaborative helpers in manufacturing and everyday life.

The researchers used machine learning to learn a person's preference based on their sequence of actions in a small assembly task, such as assembling parts for a simple model airplane.

The system could predict human actions with 82% accuracy, potentially saving time and lowering the risk of injury or accidents in industrial settings.

♾️ Meta Unleashes SAM, an AI Model That Can Spot Hidden Figures in Images

Meta, the company that owns Facebook, has released the Segment Anything Model (SAM), an artificial intelligence (AI) model that can identify individual objects within images and videos even if it has not been trained on them.

SAM can identify items in images by clicking on them or typing text prompts, and it can draw boxes around the things it finds.

Meta intends to use generative AI to create new content for its apps in the future. The SAM model and dataset are free to use under a non-commercial license.

SAM's advanced capabilities result from millions of images and masks collected using a model-in-the-loop "data engine."

The researchers used SAM and its data to annotate images and update the model interactively. This cycle was repeated several times to improve the model and the dataset.

You can try it here and read their paper.

⚠️OpenAI Takes the High Road in AI Safety with Rigorous Measures

OpenAI has outlined the steps to ensure its AI systems' safety while acknowledging potential risks.

The company rigorously tests systems, solicits external expert feedback, employs reinforcement learning, and develops safety and monitoring systems. It also collaborates with governments to create regulations.

OpenAI gradually makes its AI tools available to a more extensive user base, constantly improving them as it learns how people use them.

Regarding children, safety is a priority, with content flagged as disallowed by GPT-4 down 82% compared to its predecessor and image tools identifying Child Sexual Abuse Material.

Read more on their blog.

⬇️ Am AI the Drama?: From Optimism to Pessimism

According to the Future of Life Institute (FLI), the race to develop and deploy influential digital minds is out of control, and no one can understand, predict, or control them.

FLI called for an immediate pause on advanced AI research, attracting thousands of signatures from prominent figures in the field. However, critics argue that the group focused on the wrong concerns, such as the potential loss of human civilization to computer superintelligence, while ignoring the danger of large-scale misinformation dissemination via chatbots.

OpenAI and Meta approach the big questions surrounding AI in different ways, with OpenAI limiting access to its most advanced tools and the latter allowing researchers to probe its AI models for implicit biases and other flaws.

In other industries, potentially dangerous products are only released after private actors have satisfied public agencies that they are safe.

💬 Customized Responses in Seconds

QuickReply uses artificial intelligence (AI) to evaluate incoming messages and generate relevant, personalized responses, saving time and effort. It provides appropriate and effective responses by applying powerful algorithms to interpret message content.

Learn more about the tool here.

👒 Sense and Semantics: How Would Jane Austen Explain NLP?

Natural Language Processing (NLP) is the clever use of machines to understand the words humans use to communicate.

It's like a fancy new way of teaching computers to understand and even speak our language back to us, similar to Mr. Darcy learning to speak more warmly to Miss Bennet.

With NLP, computers can now read and analyze books, letters, and other texts, allowing us to learn more about how people talk and write, just as we might explore Emma Woodhouse and Mr. Knightley's witty conversations.

It's a genuinely remarkable technological advancement that has the potential to transform our understanding of language and communication.

🐦 #SocialScene

👔 Expert advice of the day

While AI will be growing adoption to enhance our collective user experience at scale, it will be balanced with appropriate human intervention.

🧠 #MindsetMaze

You never really learn much from hearing yourself speak. ― George Clooney

🤪 Meme of the day

🔥 TRENDING TOOLS 🔥

• 🚚 DigitalFirst.ai use AI to create and implement your marketing strategy quickly.

• 📈 VWO is an A/B testing and personalization platform that helps marketers optimize their websites and increase conversion rates.

• 💻 Botowski is an AI content generator that can facilitate high-quality content for copywriters, business owners, and others.

If you liked this edition, please share it with a friend or two! You could get some exclusive content in your inbox or even some merch very soon👇🏾

And that's a wrap!

Did we miss anything? Or want to say hey? Hit reply - I'd love to hear from you! You can

• 🐦 Find us on Twitter

• 💬 Huzzah on LinkedIn

• 📱 Slide into our DMs on Instagram

And if you haven’t already, sign up to get this in your inbox daily.