This is your daily dose of ☕️ Techpresso, where you get the latest tech news of the day effortlessly.
Here's the latest tech news from the last 24 hours:
🚀 Amazon's first internet satellite launch was a success
📖 AI learns to ignore copyrighted content through Harry Potter
🏛 Alibaba closes social science institute that Jack Ma once hoped would exist for 300 years
📸 Getty Images introduces 'socially responsible' AI for artists
🤖 AI chatbots are only useful if you think they are, scientists find
🚀 Amazon's first internet satellite launch was a successLINK
Amazon's first two Project Kuiper satellites were successfully launched into low Earth orbit via United Launch Alliance's Atlas V rocket.
This launch, named Protoflight, serves as a critical learning opportunity for Amazon, enabling the collection of off-world data and testing satellite processing, launch, and mission operations.
The eventual aim of Project Kuiper is to deliver fast, affordable broadband to unserved and underserved communities worldwide, with plans for more than 3,200 satellite launches over the next six years.
📖 AI learns to ignore copyrighted content through Harry PotterLINK
Researchers from Microsoft have developed a method to make large language models (LLMs) like OpenAI’s ChatGPT and Meta’s Llama 2 forget specific information such as copyrighted materials from Harry Potter series.
The technique involves identifying related tokens from the specific data, replacing unique expressions with generic ones and fine-tuning the model on these alternative predictions, effectively making the model "forget" the original content.
This proof-of-concept technique would help in creating more responsible, adaptable, and legally compliant language models by selectively forgetting copyrighted or unwanted information, though further research is needed for more general and robust application.
🏛 Alibaba closes social science institute that Jack Ma once hoped would exist for 300 yearsLINK
Alibaba Group has shut down its Luohan Academy, a social science research institute established in 2018, as part of a broad business restructuring.
Parts of the academy's functions, such as the environmental, social and governance research team, have been kept within Alibaba.
The closure comes as Alibaba shifts focus towards "users first" and "artificial intelligence-driven" themes, under new CEO Eddie Wu Yongming.
📸 Getty Images introduces 'socially responsible' AI for artistsLINK
Getty Images CEO, Craig Peters, has developed an AI tool that uses only licensed data, rewarding creators the more the tool is used, in an attempt to respect artists' copyrights.
Unlike other AI image generators, Getty's tool is trained exclusively on images that it owns the rights to, ensuring that there are no copyright risks involved.
If there are any lawsuits over AI images generated by Getty's tool, the company will cover any legal costs, demonstrating their confidence in this new method of respecting intellectual property.
🤖 AI chatbots are only useful if you think they are, scientists findLINK
An MIT study found that users' preconceived expectations of an AI heavily influence their reported experiences, with subjects having the types of interaction they were primed to have.
This research links users' mental models - born from cultural background and personal beliefs - to the effectiveness of AI, suggesting a user's perception greatly impacts AI interaction.
The study’s authors assert that AI developers can significantly influence user experience, highlighting the need to address not only AI biases but also users’ biases when interacting with the system.
Other news you might like
Amazon Luna's users can now buy select Ubisoft games directly on the platform, play offline on their PC or stream on Luna-enabled devices.LINK
Cheap Android TV streaming devices are coming preinstalled with unremovable malware, unknowingly turning thousands of users into nodes in a large-scale fraud scheme.LINK
"PLD Space, a Spanish firm, successfully conducted Europe's first fully private rocket launch, offering promising strides for the region's space endeavors."LINK
Synapse, a banking-service startup, laid off 40% of its staff due to challenging macroeconomic conditions and loss of a key client, Mercury.LINK
Erratic use of AI image generators continues with users manipulating tools like Meta's Llama 2 and Microsoft's Image Creator to generate inappropriate and offensive content, highlighting the failings of AI safeguards against misuse.LINK
Scientists developed an AI-based test that discerns biological from non-biological samples with 90% accuracy, aiding in the search for extraterrestrial life.LINK
Want to get the latest news differently? Find us on:
See you tomorrow for a new dose of ☕️ Techpresso!