• Data Pragmatist
  • Posts
  • OpenAI to launch custom GPT;AI Might Actually Enforce All Our Stupid Laws

OpenAI to launch custom GPT;AI Might Actually Enforce All Our Stupid Laws

Tech TidBits for today

🤖 Google wrote a ‘Robot Constitution’ to make sure its new AI droids won’t kill us.

🛍️ OpenAI to launch custom GPT store next week.

🤖 AI Might Actually Enforce All Our Stupid Laws, Expert Warns.

🔍 Google Bard Advanced leak hints at imminent launch for ChatGPT rival.

🛰️ AI and satellites expose 75% of fish industry ‘ghost fleets’ plundering oceans.

🤖 Google wrote a ‘Robot Constitution’ to make sure its new AI droids won’t kill us. Link

  • Google's DeepMind team has introduced a data gathering system, AutoRT, equipped with a Robot Constitution inspired by Isaac Asimov’s Three Laws of Robotics, designed to help robots understand their environment and make safer decisions by avoiding tasks involving humans and dangerous objects.

  • AutoRT, using visual and language models, performed over 77,000 tasks in trials with 53 robots, featuring safety measures like auto-stop and a kill switch.

  • Alongside AutoRT, DeepMind has developed additional technologies such as SARA-RT for improved accuracy and RT-Trajectory for enhanced physical task performance.

🛍️ OpenAI to launch custom GPT store next week. Link

  • OpenAI's GPT Store, enabling users to share and sell custom AI agents, is set to launch next week.

  • The platform targets ChatGPT Plus and enterprise subscribers, allowing them to build and monetize specialized ChatGPT models.

  • Although its launch was postponed from November, OpenAI is preparing GPT Builders for the upcoming release.

🤖 AI Might Actually Enforce All Our Stupid Laws, Expert Warns. Link

  • Eliezer Yudowsky warns about AI enforcing complex legal codes, threatening individuals with police reports and lawsuits for non-compliance.

  • Yudowsky highlights issues with existing AI lawyers and law enforcement tools, citing problems with facial recognition in policing.

  • He notes the distinction between AI struggles in creative tasks versus potential effectiveness in managing strict legal frameworks.

  • Yudowsky dismisses the idea of AI simplifying laws, arguing that if humans could achieve it, they would have done so already. He emphasizes the risks of AI rigidly enforcing all rules.

🔍 Google Bard Advanced leak hints at imminent launch for ChatGPT rival. Link

  • Google Bard Advanced, with exclusive features like high-level math and reasoning, is hinted to launch soon, possibly bundled with a Google One subscription.

  • Leaked information suggests new Bard features, including custom bot creation and specialized tools for brainstorming and managing tasks.

  • The exact Google One tier required for Bard Advanced access and its pricing remain undisclosed, but speculation points to the Premium plan.

🛰️ AI and satellites expose 75% of fish industry ‘ghost fleets’ plundering oceans. Link

  • A study using satellite imagery and machine learning uncovered that up to 76% of global industrial fishing vessels aren't publicly tracked, suggesting widespread unreported fishing.

  • Researchers created a global map of maritime activities, revealing concentrated vessel activity with Asia accounting for the majority, and highlighted underreporting of industrial activities at sea.

  • The growing 'blue economy' is valued at trillions but poses environmental risks, with a significant portion of fish stocks overexploited and marine habitats lost due to industrialization.