Each week, we bring you 5 stories that resonated the most in our internal Slack channel #AI-news. We write the newsletter using various AI tools because we're an AI company and our marketing wants to move with the times too. π
Today you're reading the 41st issue in a row.
#1
Pika 1.5 π₯: AI video generator, with physics-defying effects π₯π§.
Pika Labs has just released a new version of the Pika 1.5 AI video generator with new special effects called "Pikaffects" π₯.Β
These effects allow users to manipulate video objects in ways that aren't quite possible in real life - like explode π₯, melt π« , or crush objects with a simple click. With the "cake-ify" π effect, you can even turn objects into realistic cakes, adding a playful and unexpected atmosphere to your video π.
Pika 1.5 brings something fresh to the world of AI videos, as most of the competition is now focusing on the exact opposite - improving realism and camera movements π¬. It gives creators more freedom to create fun and surprising effects, setting themselves apart in a crowded market π.Β
We think it's super fun, how about you? Check out the link below to check out Piku 1.5 π!Β
Β
#2
OpenAI comes with Realtime API π€: Real-time voice interactions with AI π€ποΈ.
OpenAI has just released the Realtime API, which allows developers to create real-time voice interactions with AI using six new voices π€. These voices are specifically created for this API and offer super fast responses β‘, almost like you're talking to a real person π£οΈ.
In the demo, OpenAI demonstrated an AI assistant for trip planning πΊοΈ that not only responded with voice ποΈ, but also marked restaurants on a map at a same time. While it can't make direct calls to restaurants like Google Duo π, it can use Twilio to manage phone calls βοΈ (Twilio is a platform that provides APIs for communicating via voice, SMS and video π±). However, developers need to ensure that the people who are on the other end of these calls know they are talking to an AI π€, a step that may soon be required by California law π.
Thanks to the Realtime API, OpenAI allows you to converse with AI π€ as if you were talking to a real person π§βπ», all with lightning fast responses β‘.
Official press release on OpenAI.com
Β
#3
Researchers uncover 303 new geoglyphs in the Nazca Desert using artificial intelligence π€π€―.
In an exciting combination of archaeology and cutting-edge technology, artificial intelligence has helped uncover 303 new geoglyphs in Peru's Nazca Desert - almost as many as have been found in the entire last century π€―.
These huge ancient patterns, carved into the desert more than 2,000 years ago, have been messing with scientists' heads for decades π΅βπ« Using IBM's Watson AI system, the researchers were able to scan huge amounts of high-resolution satellite and aerial imagery π‘ that would have taken years to analyse manually π΅οΈββοΈ By searching for patterns in the soil, the AI identified more than 1,300 possible candidates for new geoglyphs, from which archaeologists confirmed 303 new patterns ποΈ
The discovery provides new insight into what these patterns may have been for - researchers think they may have shown travelers the way or marked important sites for ceremonies π€οΈπ. Combining AI and archaeology allows us to understand ancient cultures much better in a much shorter time β³ and we think it's a real bombshell π₯! What about you π?
Article on english.elpais.comΒ
Β
#4
Meta Movie Gen π€: AI editing and video generation π¬π₯
Meta has just announced the new π¬ Movie Gen, its own AI tool for creating videos from text. What makes it different is that it can edit videos that have already been shot and allows users to change objects in the videos - for example, swap the background or add new things/objects, including sounds created by AI π€ (which is new in this regard). As an example, Meta showed an AI video of a woman drinking coffee in a garden with pumpkins π, created just from her portrait photo.
Although Meta is pushing things forward with this technology β©, the company says they're not ready to launch it to the public yet due to the cost and the fact that videos take a long time to generate π. However, it is competing with other companies like OpenAI, which is also planning to launch its own AI video generator, Sora π₯.
Movie Gen's potential raises copyright and ethical concerns βοΈ, mainly because AI-generated content can blur the lines of originality. The impact of AI on the creative industries is a growing concern, with writers and actors already addressing it during the big strikes last year π.
What about you? Do you use AI video generators π€?Β
Official press release at ai.meta.com
Β
#5
Y Combinator startup PearAI under fire π¨: Controversy around copying an open-source project π₯π€.
Y Combinator-backed startup PearAI is in hot water π₯ after it launched an AI code editor that is very similar to another project, Continue.Β
Founder Duke Pan freely admitted that PearAI is essentially a "copycat" of the Continue project, which is open-sourced under the Apache license. As if that wasn't enough, PearAI first used its own license, which Pan revealed was "knocked off" by ChatGPT π€. This caused a lot of confusion because changing an open-source license undermines the principles of collaboration and transparency of the open-source community.Β
The reaction came immediately β‘. People started pointing out problems with the licenses and that PearAI looks almost identical to the original project. Pan apologized and reverted PearAI back under the original Apache license, but the damage was already done π₯.Β
This controversy raises questions about Y Combinator's decision making and about AI startups being funded without enough validation π€. This whole case shows how important it is to follow the rules of the open-source community and think twice before changing anything. What do you think about this π ?
Β
