Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 1,110

Running ChatGPT costs millions of dollars a day, which is why OpenAI, the company behind the viral natural-language processing artificial intelligence has started ChatGPT Plus, a $20/month subscription plan. But our brains are a million times more efficient than the GPUs, CPUs, and memory that make up ChatGPT’s cloud hardware. And neuromorphic computing researchers are working hard to make the miracles that big server farms in the clouds can do today much simpler and cheaper, bringing them down to the small devices in our hands, our homes, our hospitals, and our workplaces.

One of the keys: modeling computing hardware after the computing wetware in human brains.

“Inference costs far exceed training costs when deploying a model at any reasonable scale,” say Dylan Patel and Afzal Ahmad in SemiAnalysis. “In fact, the costs to inference ChatGPT exceed the training costs on a weekly basis. If ChatGPT-like LLMs are deployed into search, that represents a direct transfer of $30 billion of Google’s profit into the hands of the picks and shovels of the computing industry.”

If you run the numbers like they have, the implications are staggering.

“Deploying current ChatGPT into every search done by Google would require 512,820 A100 HGX servers with a total of 4,102,568 A100 GPUs,” they write. “The total cost of these servers and networking exceeds $100 billion of Capex alone, of which Nvidia would receive a large portion.”

Read more | >

Google employees criticized the company and CEO Sundar Pichai over the ‘botched’ launch of its ChatGPT competitor.

Googlers are talking all about the company’s announcement of its ChatGPT rival, Bard — and many aren’t happy with how things went. According to a report from CNBC, Google employees are calling the launch of the AI chatbot “rushed” and “botched” in posts across the company’s internal message boards, with many targeting CEO Sundar Pichai.

Google announced Bard earlier this week in a bid to get ahead of Microsoft, which took the wraps off of its ChatGPT-powered Bing a day later.

One employee said Google “botched” the Bard announcement.

Read more | >

Generative Pre-trained Transformer 3 (GPT-3) is an autoregressive language model that was released just a few years back. This model uses deep learning to produce human-like text, hence has immense potential. However, considering how open the market is, there are numerous alternatives available out there. Here is a list of top 10 open-source GTP-3 alternatives you should try in 2023.

Bloom

Developed by a group of over 1,000 AI researchers, Bloom is an open-source multilingual language model that is considered as the best alternative to GPT-3. It is trained on 176 billion parameters, which is a billion more than GPT-3 and required 384 graphics cards for training, each having a memory of more than 80 gigabytes.

Read more | >

❤️ Check out Weights & Biases and sign up for a free demo here: https://wandb.com/papers.

📝 The paper “Phenaki — Realistic video generation from open-domain textual descriptions” is available here:
https://phenaki.research.google/

My latest paper on simulations that look almost like reality is available for free here:
https://rdcu.be/cWPfD

Or this is the orig. Nature Physics link with clickable citations:
https://www.nature.com/articles/s41567-022-01788-5

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Aleksandr Mashrabov, Alex Balfanz, Alex Haro, Andrew Melnychuk, Benji Rabhan, Bryan Learn, B Shang, Christian Ahlin, Edward Unthank, Eric Martel, Geronimo Moralez, Gordon Child, Jace O’Brien, Jack Lukic, John Le, Jonas, Jonathan, Kenneth Davis, Klaus Busse, Kyle Davis, Lorin Atzberger, Lukas Biewald, Matthew Allen Fisher, Matthew Valle, Michael Albrecht, Michael Tedder, Nevin Spoljaric, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Rajarshi Nigam, Ramsey Elbasheer, Richard Sundvall, Steef, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Timothy Sum Hon Mun, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi.
If you wish to appear here or pick up other perks, click here: https://www.patreon.com/TwoMinutePapers.

Thumbnail background image credit: https://pixabay.com/images/id-1598620/

Continue reading “Google’s New AI: The Age of AI-Made Movies Is Here!” | >

Opera announced a new feature that will be added to its browser’s sidebar. Called ‘shorten’, the tool is a ChatGPT-powered tool that can be used to generate summaries of webpages and articles. The blog also displays a short demo video that gives us a glimpse of how ChatGPT will be integrated in the browser.

Song Lin, Co-CEO of Opera, said in the blog post, “In more than 25 years of our company’s history, we have always been at the forefront of browser innovation. Whether inventing browser tabs or providing our users with built-in access to generative AI tools, we always push the limits of what’s possible on the web. Following the mass interest in generative AI tools, we believe it’s now time for browsers to step up and become the gateway to an AI-powered web”.

Read more | >

ChatGPT, Open AI’s conversational chatbot, is currently one of the most popular tools on the internet. Designed to assist in tasks ranging from summarizing information to generating some of its own, the chatbot has trumped the likes of Tiktok and Instagram regarding daily active users. Microsoft hopes to cash into this rush by powering its Bing search engine with an advanced version of the GPT, the learning language model that runs the chatbot.

Alilbaba to launch ChatGPT rival too

Such has been the craze of ChatGPT that no tech company wants to be left out of this race. After years of hyped-up AI talk, something tangible has emerged and has the potential to knock Google off its perch or at least drastically change how it conducts its business.

Read more | >

Meta, formerly known as Facebook, today showed off a prototype of an AI system that enables people to generate or import things into a virtual world just by using voice commands. The company sees the tool, which is called “Builder Bot,” as an “exploratory concept” that shows AI’s potential for creating new worlds in the metaverse. Meta CEO Mark Zuckerberg showed off the prototype at the Meta AI: Inside the Lab event on Wednesday in a pre-recorded demo video.

In the video, Zuckerberg explained the process of building parts of a virtual world by describing them. He begins with the prompt, “let’s go to a park.” The bot then creates a 3D landscape of a park with green grass and trees. Zuckerberg then says “actually, let’s go to the beach,” after which the bot replaces the current landscape with a new one of sand and water. He then says he wants to add clouds and notes that everything is AI-generated. Zuckerberg then changes up the landscape by saying he’d rather have altocumulus clouds, which is meant to demonstrate how specific the voice commands can be.

He then points to a specific area of the water and says “let’s add an island over there,” and then the bot creates one. Zuckerberg then issues several other voice commands, such as adding trees and a picnic blanket. He also adds the sound of seagulls and whales. At one point, he even adds a hydrofoil — a nod to one of his favorite hobbies, which later turned into a meme.

Read more | >

The current media environment is filled with visual effects and video editing. As a result, as video-centric platforms have gained popularity, demand for more user-friendly and effective video editing tools has skyrocketed. However, because video data is temporal, editing in the format is still difficult and time-consuming. Modern machine learning models have shown considerable promise in enhancing editing, although techniques frequently compromise spatial detail and temporal consistency. The emergence of potent diffusion models trained on huge datasets recently caused a sharp increase in the quality and popularity of generative techniques for picture synthesis. Simple users may produce detailed pictures using text-conditioned models like DALL-E 2 and Stable Diffusion with only a text prompt as input. Latent diffusion models effectively synthesize pictures in a perceptually constrained environment. They research generative models suitable for interactive applications in video editing due to the development of diffusion models in picture synthesis. Current techniques either propagate adjustments using methodologies that calculate direct correspondences or, by finetuning on each unique video, re-pose existing picture models.

They try to avoid costly per-movie training and correspondence calculations for quick inference for every video. They suggest a content-aware video diffusion model with a configurable structure trained on a sizable dataset of paired text-image data and uncaptioned movies. They use monocular depth estimations to represent structure and pre-trained neural networks to anticipate embeddings to represent content. Their method gives several potent controls on the creative process. They first train their model, much like image synthesis models, so the inferred films’ content, such as their look or style, correspond to user-provided pictures or text cues (Fig. 1).

Figure 1: Video Synthesis With Guidance We introduce a method based on latent video diffusion models that synthesises videos (top and bottom) directed by text-or image-described content while preserving the original video’s structure (middle).

Read more | >

In Celebration of the Recent 20-Year Anniversary of Snapple’s Real Facts®, Snapple is Putting its Fact Writing into Fans’ Hands.

FRISCO, Texas, Feb. 8, 2023 /PRNewswire/ — Snapple®, the iconic beverage brand that delivers fun and flavorful teas and juice drinks, is proud to announce the launch of the Snapple fAIct Generator, an AI-powered tool that makes it easy to create facts about any topic. Celebrating 20-years of Snapple Real Facts®, facts found under every Snapple bottle cap, the Snapple fAIct Generator puts fact-creation in the hands of the brand’s fans. To help share the news of this new tool, Snapple used ChatGPT to write this press release, with some light edits to make it more Snapple-y.

Read more | >

Lurking inside your next gadget may be a chip unlike those of the past. People used to do all the complex silicon design work, but for the first time, AI is helping to build new chips for data centers, smartphones, and IoT devices. AI firm Synopsys has announced that its DSO.ai tool has successfully aided in the design of 100 chips, and it expects that upward trend to continue.

Companies like STMicroelectronics and SK Hynix have turned to Synopsys to accelerate semiconductor designs in an increasingly competitive environment. The past few years have seen demand for new chips increase while materials and costs have rocketed upward. Therefore, companies are looking for ways to get more done with less, and that’s what tools like DSO.ai are all about.

The tool can search design spaces, telling its human masters how best to arrange components to optimize power, performance, and area, or PPA as it’s often called. Among those 100 AI-assisted chip designs, companies have seen up to a 25% drop in power requirements and a 3x productivity increase for engineers. SK Hynix says a recent DSO.ai project resulted in a 15% cell area reduction and a 5% die shrink.

Read more | >