This week’s amazing tech stories from around the web (until March 11)

D-ID’s new web app gives OpenAI’s ChatGPT a face and a voice
Aisha Malik | TechCrunch
“When you open the web app on desktop or mobile, you’ll be greeted by an avatar named ‘Alice.’ You can then either type a question or click the microphone icon to say your question out loud. D-ID finds that Alice can answer almost anything. You can ask Alice to simulate a job interview or even host your family’s quiz night. … In a few weeks, the web app will allow users to create and talk to a character like Dumbledore from Harry Potter.”

Two strange ideas for a megaqubit quantum computer
Samuel K Moore | IEEE spectrum
“Experts say that quantum computers may need at least a million qubits maintained near absolute zero to perform anything computationally remarkable. But it would be impossible to connect them all via coaxial cable to control and readout electronics that work at room temperature. Computing giants like IBM, Google and Intel are hoping to solve this problem with cryogenic silicon chips that can work close to the qubits themselves. But researchers have recently proposed some more exotic solutions that could speed up the pace.”

The discovery of room-temperature superconductors is met with resistance
Charlie Wood and Zack Savitsky | quantum
“The results, published [this week] In Nature, seem to show that a conventional conductor – a solid made of hydrogen, nitrogen and the rare-earth metal lutetium – has been transformed into a pristine material capable of conducting electricity with perfect efficiency. While the announcement was greeted with enthusiasm by some scientists, others are far more cautious, pointing to the research group’s controversial history of alleged research misconduct.”

Sam Altman has invested $180 million in a company trying to delay death
Antonio Regalado | MIT Technology Review
[Altman] says he emptied his bank account to fund two other very different but equally ambitious goals: limitless energy and extended lifespan. One of those bets is on fusion energy startup Helion Energy, into which he has poured more than $375 million, he told CNBC in 2021. The other is Retro, for which Altman put checks totaling $180 million that same year US dollar shortened. ‘That is much. I basically just took all my liquid assets and put them into these two companies,” Altman says.”

Meta’s powerful AI language model leaked online – what happens now?
Jacob Vincent | The edge
“Meta has not released LLaMA as a public chatbot (although the Facebook owner Is build them as well), but as an open-source package that anyone in the AI ​​community can request access to. …However, just a week after Meta began responding to requests for access to LLaMA, the model was leaked online. A downloadable torrent of the system was released on 4chan on March 3rd and has since spread across various AI communities, sparking debate about the right way to share cutting-edge research in an era of rapid technological change.

Forget designer babies. This is how CRISPR really changes lives
Antonio Regalado | MIT Technology Review
“…there are now more than 50 experimental studies underway using gene editing in human volunteers to treat everything from cancer to HIV and blood disorders, according to a review by David Liu, a gene editing specialist at Harvard University, shared with MIT Technology Review. Most of these studies—about 40 of them—involve CRISPR, the most versatile of gene editing methods, only developed 10 years ago.”

Could the next blockbuster drug be lab rat free?
Emily Anthes | The New York Times
“…there is momentum building for non-animal approaches that could ultimately help accelerate drug development, improve patient outcomes and reduce the burden of laboratory animals, experts said. “Animals are simply surrogates for predicting what will happen inside a human,” said Nicole Kleinstreuer, director of the National Toxicology Program Interagency Center for the Evaluation of Alternative Toxicological Methods. “If we can get to a point where we actually have a fully human-relevant model,” she added, “then we won’t need the animal black box anymore.”I

This geothermal startup showed that its wells can be used like a giant underground battery
James Temple | MIT Technology Review
“The results of the first experiments … indicate that Fervo can build flexible geothermal power plants, capable of increasing or decreasing the power output as needed. Potentially more important, the system can store energy for hours or even days and release it back over similar periods of time, effectively acting like a huge and very long-lasting battery. That means the plants could shut down production as solar and wind farms ramp up, and provide a plentiful stream of clean electricity when those sources dwindle.”


Detection stays one step ahead of deepfakes – for now
Matthew Hutson | IEEE spectrum
“…while computer scientists develop better methods for algorithmically generating video, audio, images, and text—typically for more constructive purposes, such as enabling artists to manifest their visions—they also develop counter-algorithms to detect such synthetic content. Recent research shows advances in the robustness of detection, sometimes by looking beyond the subtle signatures of certain generation tools and instead leveraging underlying physical and biological cues that are difficult for AI to imitate.”

GPT-4 might just be a bloated, pointless mess
Jacob Stern | The Atlantic
Will the endless “scaling” of our current language models really bring real machine intelligence? ...the scaling debate is representative of the broader AI discourse. It feels like the vocal extremes have drowned out the majority. Either ChatGPT will completely transform our world or it’s a glorified toaster. The boosters praise their 100-proof hype, the critics respond with leaden pessimism, and the rest of us sit quietly somewhere in the middle trying to make sense of this strange new world.

Credit: Laura Skinner/Unsplash

Leave a Reply

Your email address will not be published. Required fields are marked *