[ad_1]
Hey guys, welcome to TechCrunch’s regular AI newsletter. If you want this in your inbox every Wednesday, sign up here.
It’s almost Thanksgiving, and you know what that means: stuffing your face to the point of coma. Well, that and entertaining all the obnoxious relatives I’ve managed to keep away for the rest of the year.
For those of us lucky (unlucky?) enough to be tasked with fixing this year’s feast, there’s not much time left to settle on a menu. It’s always a tough decision, what to prepare for. So, here’s a smart idea: ask a chatbot.
Yes, yes, it’s been done before — turning to artificial intelligence for desperate help on Thanksgiving. (The New York Times tried ChatGPT Recipes In 2022.) The results were generally average. But perhaps the claims are the problem.
Curiosity got the best of me. So I asked some of the most popular chatbots, ChatGPT and Claude, to come up with a Thanksgiving menu that was “so unique that it would wow any family member.” I thought this would do the trick.
Let me tell you, reader, that AI did not disappoint you.
ChatGPT recommends starting with a cocktail hour – great! – Features whipped sweet potatoes and goat cheese crostini. Meanwhile, Claude shot for the moon, suggesting an appetizer — “pumpkin soup with sage foam” — that definitely checked the “unique” box.
Does “pumpkin soup with crème fraîche” sound good? That’s what ChatGPT suggested for the appetizer, followed by a main course of miso-buttered turkey with ginger-soy glaze. “Dry-brined turkey with lavender and fennel with a honey-thyme glaze,” suggested Claude, who was again the wild card. The chatbot described it as a herbal departure from the classic roast turkey. actually.
What about the sides? ChatGPT recommended the chili cornbread and pistachio risotto. Claude said fine wines should be made to have a “wild mushroom and chestnut filling with aged sherry.”
For the big finish, both chatbots will have you stick to staple foods: pie, cheesecake, and healthy scoops of ice cream. development? Saffron flavored ice cream, and tea spiced cheesecake.
“This menu takes familiar Thanksgiving flavors and elevates them with unexpected ingredients, techniques, and combinations,” Claude writes of his creations. “Each dish tells a story and invites conversation, making a meal not just about the food, but also about shared experience and creativity.”
I can’t argue with that. But as this year’s designated chef…well, let’s just say I won’t be aiming for the Top Chef title.
news
Sora leaks from OpenAI: A group appears to have leaked access to Sora, OpenAI’s video generator, in protest of what it calls “artwashing” on the part of OpenAI.
Amazon supports Anthropic once again: Anthropic has He grew up An additional $4 billion from Amazon, and it agreed to train its leading generative AI models primarily on Amazon Web Services, Amazon’s cloud computing division.
AI application connectors: In other humanitarian news, the company has proposed a new standard, the Model Context Protocol, to connect AI assistants to the systems where the data resides.
OpenAI funds “AI ethics” research: OpenAI is pouring $1 million into a research program at Duke University to develop algorithms that can predict moral judgments in humans.
YouTube gets artificial intelligence wallpapers: YouTube’s Dream Screen short video feature, the platform’s short video format, now allows users to create AI-generated video backgrounds.
Brave adds chat AI: Search engine Brave has introduced an AI chat mode for follow-up questions based on initial queries on Brave Search, an expansion of Brave’s Answer with AI feature that provides AI-generated summaries of web searches.
Ai2 open source tullo 3: The Allen Institute for Artificial Intelligence (Ai2) has released Tülu 3, a generative AI model that can be fine-tuned and customized for a range of applications (such as solving mathematical problems).
Crusoe collects money: Crusoe Energy, a data center builder startup that will reportedly lease to Oracle, Microsoft and OpenAI, is in the process of raising $818 million, according to a Securities and Exchange Commission filing.
Topic tests and artificial intelligence summaries: Meta’s Threads has begun testing AI-generated summaries of what people are discussing on the platform, taking a page from competitor X.
Research paper of the week
DeepMind, Google’s AI research organization, has developed a new AI system called AlphaQubit that it claims can pinpoint errors inside quantum computers.
Quantum computers are likely to be much more powerful than classical machines for certain workloads. But they are also more susceptible to “noise” or general errors.
AlphaQubit identifies these errors so they can be mitigated and corrected, helping make quantum computers more reliable.
It’s not a flawless system, though. Google acknowledges a mail AlphaQubit is very slow at real-time debugging, and is not particularly efficient at handling data. The company says work is underway on improved versions.
Model of the week
Runway, a startup that builds AI tools for content creators, has released a new image generation model that the company claims offers better stylistic control than most.
Named FramesThe model, which is being slowly rolled out to users of Runway’s Gen-3 Alpha video generator, can reliably create images that stay true to a particular aesthetic, Runway says.
Now, it’s worth noting that Runway may be fast and not bound by copyright rules. A 404 Media Report Earlier this year, the company suggested it would delete YouTube footage from channels affiliated with Disney and creators like MKBHD without obtaining permission to train its models.
When reached for comment, a Runway spokesperson declined to reveal the source of Frames’ training data.
Like many generative AI companies, Runway maintains that its data collection practices are protected by fair use. This theory is being tested in a number of courtroom battles, including a class-action lawsuit filed against Runway and several of its rival art makers.
Grab a bag
Nvidia has unveiled a model it calls “the world’s most flexible audio machine.”
The chip giant’s model, dubbed Fugatto, can create a mix of music, voices and sounds from a text description and a set of audio files. For example, Fugatto can create a musical snippet based on a prompt, remove or add instruments from/to a song, and change the tone or emotion of a vocal performance.
Nvidia claims that Fugatto, trained on millions of publicly licensed sounds and songs, can also create objects that don’t exist in the real world.
“For example, Fugatto can make a trumpet bark or a saxophone meow,” the company says. books In a blog post. “Through fine-tuning and small amounts of singing data, researchers found that it could handle tasks for which it had not been (trained), such as generating a high-quality singing voice from a text prompt.”
Nvidia did not release Fugatto for fear of it being abused. but According to Reuters said that the company is studying how to launch the model “responsibly” if it is made available.
[ad_2]