Home Blog Page 83

Life-size Titanic digital twin unveiled

0

The Fate of the Titanic: A New Dimension of Understanding

A Digital Twin of the Wreckage

The fate of the Titanic has been a source of fascination for over a century. There have been expeditions to the wreckage in the Atlantic, but it’s so dark 4,000 meters (over 13,000 feet) under the sea that individual fotos and video provide a limited view.

Transforming Our Understanding

Now our understanding of how the Titanic sank could be transformed thanks to the creation of a digital twin of the wreckage based on 3D scans. The full-scale replica features in a new National Geographic documentary: Titanic: The Digital Resurrection.

The Model

The model captures the wreck in incredible detail, from the fractured hull plates to debris on the seabed, and it allows scientists to explore the ship at its real size to learn more about how it sank after hitting an iceberg in 1912.

New Discoveries

The replica reveals a series of small, A4 paper-sized punctures along the hull, which challenges the prevailing idea that the iceberg tore one large gash. The discovery suggests that water slowly overwhelmed the ship’s design limits.

The Wreckage

The Titanic was described as unsinkable because it was designed to stay afloat even if four of its watertight compartments flooded. But a new simulation suggests that punctures were inflicted across six compartments. Water coming slowly into all of those holes would have eventually flooded the compartments.

The Digital Twin

The wreck was mapped using underwater robots called virtual remotely operated vehicles, or vROVs. Information can now be combined with a structural model created from blueprints along with measurements and information on speed, direction, and position using numerical algorithms to explore how the Titanic may have sunk.

New Details

More new details have already been identified, including a porthole that may have been smashed by the iceberg. The scan also provides a view of a boiler room showing some of the boilers are concave, potentially confirming eye-witness testimony from the time that suggested that engineers continued to work to keep the lights on as the ship sank.

Conclusion

The creation of a digital twin of the Titanic wreckage is a groundbreaking development that has the potential to transform our understanding of this tragic event. The model provides a detailed and accurate representation of the ship, allowing scientists to explore its design and construction in unprecedented detail.

Frequently Asked Questions

Q: What is a digital twin?
A: A digital twin is a virtual replica of a physical object, in this case, the Titanic wreckage, created using 3D scans and other data.

Q: How was the digital twin created?
A: The digital twin was created using 3D scans of the wreckage, which were then combined with a structural model created from blueprints along with measurements and information on speed, direction, and position using numerical algorithms.

Q: What are the implications of the digital twin?
A: The digital twin has the potential to transform our understanding of the Titanic’s fate, providing new insights into the ship’s design and construction, and how it sank.

Q: When will the documentary be released?
A: The documentary, Titanic: The Digital Resurrection, will be released on National Geographic US on 13 April and National Geographic UK on 15 April.

Claude Subscription

0

Anthropic Launches Expensive Subscription Plan for AI Chatbot Claude: Max

Anthropic announced on Wednesday that it’s launching a new, very expensive subscription plan for its AI chatbot Claude: Max. An answer to OpenAI’s $200-a-month ChatGPT Pro tier, Max comes with higher usage limits than Anthropic’s $20-per-month Claude Pro subscription as well as priority access to the company’s newest AI models and features.

Two Flavors of Max

A bit confusingly, Max comes in two flavors with different price points and usage limits. There’s a $100-per-month Max tier with 5x higher rate limits than Claude Pro and a $200-per-month Max option with 20x higher rate limits.

Frontier AI Model Developers Seeking New Revenue Streams

Frontier AI model developers are looking for new ways to increase revenue — and selling power users an expensive AI subscription seems like a promising way to do it. Just two months after launching ChatGPT Pro, OpenAI reportedly told investors its annualized revenue grew by $300 million. Should Anthropic’s new Max plan be similarly successful, it could give the company a much-needed boost.

Anthropic’s Product Lead on Future Plans

Anthropic’s product lead Scott White told TechCrunch in an interview that the company is not ruling out launching even pricier subscriptions someday. Notably, Anthropic still doesn’t have an unlimited usage plan, which OpenAI offers with ChatGPT Pro.

When asked whether Anthropic would ever consider launching a $500-per-month Claude plan, White said, “We’ll always keep a number of exploratory options available to us.” User feedback will continue to guide Anthropic’s product roadmap, he added.

New Revenue Channels

To meet the massive costs of developing frontier AI models, Anthropic is exploring various new revenue channels, one of which is Claude for Education. Claude for Education targets university customers, providing specific capabilities and benefits for colleges.

Demand for Claude Products

Anthropic declined to tell TechCrunch how many Claude subscriptions it has sold to date. However, White said the company’s new Claude 3.7 Sonnet AI model has created “a lot of demand” for its products. Claude 3.7 Sonnet is Anthropic’s first reasoning model, a type of model that uses more computing power than traditional models to answer certain questions more reliably.

Conclusion

Anthropic’s new Max subscription plan aims to capitalize on the growing demand for AI-powered chatbots and other frontier AI models. With its focus on providing high-end features and priority access to new models and features, Max is poised to attract power users and enterprises looking for advanced AI capabilities.

FAQs

What is the difference between Claude Pro and Max?

Max comes with higher usage limits and priority access to new models and features compared to Claude Pro.

How many flavors of Max are available?

Max is available in two flavors: a $100-per-month option with 5x higher rate limits than Claude Pro and a $200-per-month option with 20x higher rate limits.

Will Anthropic launch even pricier subscriptions in the future?

According to Anthropic’s product lead Scott White, the company is open to exploring even pricier subscription options in the future, but will continue to prioritize user feedback in shaping its product roadmap.

What is Claude for Education?

Claude for Education is a new revenue channel for Anthropic, targeting university customers and providing specific capabilities and benefits for colleges.

How many Claude subscriptions has Anthropic sold to date?

Anthropic declined to disclose the number of Claude subscriptions it has sold to date, but noted that its new Claude 3.7 Sonnet AI model has created “a lot of demand” for its products.

Boosting AI Trust with Web3 Tech

The Promise of AI

The promise of AI is that it’ll make all of our lives easier. And with great convenience comes the potential for serious profit. The United Nations thinks AI could be a $4.8 trillion global market by 2033 – about as big as the German economy.

But forget about 2033: in the here and now, AI is already fueling transformation in industries as diverse as financial services, manufacturing, healthcare, marketing, agriculture, and e-commerce. Whether it’s autonomous algorithmic ‘agents’ managing your investment portfolio or AI diagnostics systems detecting diseases early, AI is fundamentally changing how we live and work.

But cynicism is snowballing around AI – we’ve seen Terminator 2 enough times to be extremely wary. The question worth asking, then, is how do we ensure trust as AI integrates deeper into our everyday lives?

Transparency: Opening the AI Black Box

For all their impressive capabilities, AI algorithms are often opaque, leaving users ignorant of how decisions are reached. Is that AI-powered loan request being denied because of your credit score – or due to an undisclosed company bias? Without transparency, AI can pursue its owner’s goals, or that of its owner, while the user remains unaware, still believing it’s doing their bidding.

One promising solution would be to put the processes on the blockchain, making algorithms verifiable and auditable by anyone. This is where Web3 tech comes in. We’re already seeing startups explore the possibilities. Space and Time (SxT), an outfit backed by Microsoft, offers tamper-proof data feeds consisting of a verifiable compute layer, so SxT can ensure that the information on which AI relies is real, accurate, and untainted by a single entity.

Proving AI Can Be Trusted

Trust isn’t a one-and-done deal; it’s earned over time, analogous to a restaurant maintaining standards to retain its Michelin star. AI systems must be assessed continually for performance and safety, especially in high-stakes domains like healthcare or autonomous driving. A second-rate AI prescribing the wrong medicines or hitting a pedestrian is more than a glitch, it’s a catastrophe.

This is the beauty of open-source models and on-chain verification via using immutable ledgers, with built-in privacy protections assured by the use of cryptography like Zero-Knowledge Proofs (ZKPs). Trust isn’t the only consideration, however: Users must know what AI can and can’t do, to set their expectations realistically. If a user believes AI is infallible, they’re more likely to trust flawed output.

Compliance and Accountability

As with cryptocurrency, the word compliance comes often when discussing AI. AI doesn’t get a pass under the law and various regulations. How should a faceless algorithm be held accountable? The answer may lie in the modular blockchain protocol Cartesi, which ensures AI inference happens on-chain.

Cartesi’s virtual machine lets developers run standard AI libraries – like TensorFlow, PyTorch, and Llama.cpp – in a decentralised execution environment, making it suitable for on-chain AI development. In other words, a blend of blockchain transparency and computational AI.

Trust Through Decentralisation

The UN’s recent Technology and Innovation Report shows that while AI promises prosperity and innovation, its development risks “deepening global divides.” Decentralisation could be the answer, one that helps AI scale and instils trust in what’s under the hood.

Conclusion

In conclusion, ensuring trust in AI is crucial for its widespread adoption. Transparency, compliance, and accountability are key components in building trust. Decentralisation can be a game-changer in this regard, allowing for on-chain verification and immutable ledgers. By combining these elements, we can create a trustworthy AI ecosystem that benefits humanity.

FAQs

What is the impact of AI on our lives?

AI is transforming industries and changing how we live and work. It has the potential to make our lives easier, but also raises concerns about job security, privacy, and accountability.

How can we ensure trust in AI?

Ensuring trust in AI requires transparency, compliance, and accountability. Decentralisation can be a key component in building trust, allowing for on-chain verification and immutable ledgers.

What is Web3 tech and how does it relate to AI?

Web3 tech is the next generation of the internet, built on blockchain and decentralisation. It has the potential to revolutionise AI by providing transparent, verifiable, and auditable processes.

What is Cartesi and how does it relate to AI?

Cartesi is a modular blockchain protocol that allows for on-chain AI development. It enables developers to run standard AI libraries in a decentralised execution environment, making it suitable for on-chain AI development.

Photoshop’s First AI Agent Unleashed

Adobe’s Agentic AI: Taking AI Assistance to the Next Level

Adobe Photoshop

Adobe Photoshop already has some agentic adjacent features, such as the Distraction Removal feature, which allows users to remove background distractions like poles and wires with one click. At Adobe Max the company will build on this groundwork, unveiling a new Action Panel, the first creative agent in Photoshop.

Adobe says that with the Action Panel Photoshop will be able to analyze your image and then recommend context-aware edits. After suggesting the edit, it will be able to carry out the reaction for you with one click. Users will also be able to use natural language or normal, everyday language to access more than 1,000 one-click actions that can carry out the intended task for you.

According to Adobe this will not only help speed up the process but also ensure that users learn and familiarize themselves with the applications and learn how to use the tools better by providing feedback and suggestions.

Adobe Premiere Pro

Just last week Adobe added an AI feature, Media Intelligence, in Premiere Pro that can help you filter through all the prompts using natural language that describes the scene or angle, or includes a spoken word or an element from the metadata, such as the shoot date or the camera type.

Taking it a step further, Adobe is working on AI agents that can use that advanced understanding of your media to take actions for you, such as developing a rough cut. Adobe says that oftentimes the most difficult part of starting a project is finding all of the right shots and piecing them together. This agent will be able to help sort all of that content for you and place it on your timeline with your direction so you can jump straight into the editing.

Once you get started, the agent can provide suggestions and even help you learn how to perform complex tasks efficiently, expanding your skillset. Agents that can help assist with color grading and mix audio are also part of Adobe’s greater vision for agentic AI in Premiere Pro in the future.

Adobe Acrobat

Adobe Acrobat is one of the company’s most widely known tools, with people from all different professions using it to view and interact with digital documents. The app already has an Acrobat AI Assistant, a feature that answers information about the documents and generates quick summaries that Adobe says is a first step into agentic AI.

The company is working to build on that, bringing more agentic AI capabilities that enhance how users interact with their documents even further. For example, in the coming months there will be a feature that allows users to create custom agents with different roles, such as a research or sales assistant, or even a tutor. These agents will be able to analyze the documents, use reasoning to complete the task assigned to them, and make suggestions about further areas to explore, according to the release.

Adobe Express

Adobe Express is a content creation tool that caters to everyone, from beginners to experienced creative professionals, where people can create videos, PDFs, graphics, and more. Agentic AI in Adobe Express is meant to make the aforementioned content creation process more "intuitive and efficient," according to the blog post.

Specifically, Adobe is building an agent that can function as a creative partner, giving users support in carrying out their intended project, even if they are not familiar with what tools to use or how to get there. Adobe shares the example of an Express user asking Express to create and improve a design.

Conclusion

Adobe’s agentic AI plans will revolutionize the way users interact with their applications, making tasks faster and more efficient. With AI agents that can analyze, reason, and take action, users will be able to focus on what they love while the AI handles the rest.

Frequently Asked Questions

Q: What is agentic AI?
A: Agentic AI refers to AI that can carry out tasks for users, taking AI assistance to the next level.

Q: Which Adobe applications will have agentic AI capabilities?
A: Photoshop, Premiere Pro, Acrobat, and Creative Cloud will have agentic AI capabilities.

Q: What is the Action Panel in Photoshop?
A: The Action Panel is the first creative agent in Photoshop that can analyze your image and recommend context-aware edits, and then carry out the edit with one click.

Q: What is the Media Intelligence feature in Premiere Pro?
A: The Media Intelligence feature helps you filter through all the prompts using natural language that describes the scene or angle, or includes a spoken word or an element from the metadata.

Q: What is the Acrobat AI Assistant?
A: The Acrobat AI Assistant answers information about the documents and generates quick summaries, which is a first step into agentic AI.

Revopoint’s Markerless 3D Scanner

0

Revopoint Launches Trackit 3D Tracking Scanner: A Game-Changer in Industrial-Grade Accuracy

Revopoint, a leading manufacturer of 3D scanners, has announced the launch of the Trackit 3D Tracking Scanner, a cutting-edge optical tracking scanner designed to revolutionize the way we capture and analyze 3D data.

Eliminating Markers: A Breakthrough in Workflow Efficiency

The Revopoint Trackit 3D Tracking Scanner is designed to enhance workflow efficiency by eliminating the need for markers, a tedious and time-consuming process in traditional 3D scanning solutions. The advanced optical tracking system tracks the scanner’s targets and provides real-time error compensation, ensuring stability and accuracy.

Advanced Features and Capabilities

The Trackit boasts a one-piece carbon fiber frame that eliminates potential errors caused by frame deformation, providing durability for demanding industrial environments. The dual-mode scanning mode offers:

  • 30 blue laser cross-lines for rapidly capturing surface areas, including dark and shiny ones
  • Single-line mode for deep holes and crevices

The dual-camera optical tracking base station can easily be repositioned to ensure full surface coverage of a large capture area, making it an effective solution for scanning entire vehicles or industrial machinery.

Applications and Benefits

The Revopoint Trackit 3D Tracking Scanner can handle a wide range of challenging applications, including:

  • Complex-structured assemblies
  • Highly reflective metallic surfaces
  • Digital archival of art installations
  • Mold making/injection processes

Key Benefits

  • Eliminates the need for markers, reducing workflow time and increasing efficiency
  • Advanced optical tracking system for real-time error compensation and stability
  • One-piece carbon fiber frame for durability and accuracy
  • Dual-mode scanning for flexibility and adaptability

Pricing and Availability

For a limited time, you can sign up for a 35% off early-bird offer at Revopoint. VIP benefits include a 3-year warranty, free air shipping, and priority delivery.

Comparison with Other 3D Scanning Solutions

See below for pricing on other 3D scanning solutions:

Conclusion

The Revopoint Trackit 3D Tracking Scanner is a significant innovation in the field of industrial-grade 3D scanning. Its advanced features and capabilities make it an ideal solution for a wide range of applications, from fine-detail scanning to complex assemblies. With its focus on workflow efficiency and accuracy, the Trackit is set to revolutionize the way we capture and analyze 3D data.

Frequently Asked Questions

Q: What is the Revopoint Trackit 3D Tracking Scanner?
A: The Revopoint Trackit 3D Tracking Scanner is an optical tracking scanner designed for fine-detail scanning and industrial-grade accuracy.

Q: What are the key features of the Trackit?
A: The Trackit features advanced optical tracking, a one-piece carbon fiber frame, dual-mode scanning, and a dual-camera optical tracking base station.

Q: What are the applications of the Trackit?
A: The Trackit can be used for a wide range of applications, including complex-structured assemblies, highly reflective metallic surfaces, digital archival of art installations, and mold making/injection processes.

Q: What is the pricing of the Trackit?
A: The pricing of the Trackit is available upon request. For a limited time, you can sign up for a 35% off early-bird offer at Revopoint.

Q: What are the benefits of the Trackit?
A: The benefits of the Trackit include eliminating the need for markers, advanced optical tracking, durability, and flexibility.

A Hat in Time

0

BAFTA Game Awards 2025: Winners and Surprises

Astro Bot Dominates, But Neva Steals the Show

The 2025 BAFTA Game Awards winners have been announced, and it’s no surprise that Astro Bot from Sony’s Team Asobi had a great night, scooping five awards including Best Game, Best Animation, and Best Game Design. However, many are intrigued by the success of a lesser-known title, Nomada Studio’s graceful side-scrolling platformer Neva.

Neva’s Well-Deserved Accolade

Neva won the Best Artistic Achievement award, beating out Astro Bot, Black Myth Wukong, Still Wakes the Deep, and Senua’s Saga: Hellblade II. It was a well-deserved accolade for one of the most beautiful indie games we’ve seen. Neva follows a young woman named Alba who must travel with her wolf companion Neva across four seasons in a world corrupted by darkness.

Gameplay and Art Style

The game is very much about the bond between Alba and Neva, who must fight battles and solve puzzles through four seasons. The gorgeous painterly art style surpasses even the same developer’s 2018 game Gris, displaying a clear Miyazaki influence, which also comes across in the game’s themes.

Other Winners

Meanwhile, Senua’s Saga: Hellblade II picked up the award for Technical Achievement, using Unreal Engine 5 to bring to life striking landscape design, photo-real character models, and evocative lighting. Still Wakes the Deep, the beautifully atmospheric horror from Secret Mode won Best New Intellectual Property. Helldivers 2 won two awards for Multiplayer and Music. The wonderfully quirky Thank Goodness You’re Here was named Best British Game, and LocalThunk’s Balatro won best debut game.

Conclusion

The BAFTA Game Awards winners are chosen by BAFTA members and people highly regarded in the gaming industry who vote for their top choices in each category. The results are a testament to the incredible talent and creativity in the gaming industry. What do you think of the results? Have you played Neva and do you think it deserved its win? Let us know in the comments below.

FAQs

Q: Who won the most awards at the BAFTA Game Awards 2025?
A: Astro Bot from Sony’s Team Asobi won five awards, including Best Game, Best Animation, and Best Game Design.

Q: What was the most surprising win at the BAFTA Game Awards 2025?
A: Neva, a lesser-known indie game, won the Best Artistic Achievement award, beating out more well-known titles.

Q: What is Neva about?
A: Neva is a side-scrolling platformer that follows a young woman named Alba who must travel with her wolf companion Neva across four seasons in a world corrupted by darkness.

Q: What is the BAFTA Game Awards?
A: The BAFTA Game Awards are an annual awards ceremony that recognizes excellence in video games, chosen by BAFTA members and people highly regarded in the gaming industry.

Samsung’s The Frame Pro was Never Going to be Cheap

0

The Samsung Frame Pro: A High-End Smart TV with a Price Tag to Match

Introduction

Samsung made waves at CES earlier this year when it announced The Frame Pro. We eventually learned that the company was over-selling the TV’s "Mini LED" technology, but the new model does come with legitimate upgrades like a Wireless One Connect Box, meaning the only cable running to the TV itself will be for power, and a faster 144Hz maximum refresh rate. Now Samsung has announced how much The Frame Pro will cost when it starts shipping in the coming weeks.

Pricing and Availability

You’ll only be able to get it in three sizes, with 65-inch ($2,199), 75-inch ($3,199), and 85-inch ($4,299) models to pick between. The standard 2025 Frame tops out at 65 inches, so Samsung is dividing the line based on just how big you want to go. If you’re curious about what the premium is for going "Pro," the 65-inch regular Frame costs $1,799. So it’s definitely more expensive — but not outlandishly so. No matter which Frame you buy, you’re always paying a fair amount extra for the look and aesthetic.

Features and Upgrades

The Wireless One Connect box works at distances of up to 30 feet, and as usual, Samsung offers a selection of customizable bezels to help the TV blend in with your other decor. As with recent models of The Frame, The Frame Pro has a matte, anti-glare display that helps the artwork it’s showcasing look more realistic and less like a lit-up screen.

The Neo QLED Panel

But I remain somewhat skeptical about how much of an upgrade the Neo QLED panel will be. Samsung is using the "Mini LED" term pretty loosely here; whereas most Mini LED TVs have a huge array of small LEDs behind the screen to precisely control contrast and brightness, The Frame Pro puts all of its Mini LEDs at the bottom of the display, and they fire upward.

The S95F Series

So far I’m not seeing any tariff penalty here: these prices closely align with last year’s S95D series. Samsung says the S95F is its brightest OLED TV yet, and the Glare Free display treatment is even better at combatting reflections — with virtually no tradeoffs this time. As ever, the company says the S95F delivers better AI-powered upscaling than ever, and it can game at up to 165Hz. You won’t find Dolby Vision here: Samsung just refuses to change its tune on that, but at least streaming services are starting to get more serious about HDR10 Plus. The S95 line is available to order now with a larger 85-incher coming "soon" according to Samsung.

Conclusion

The Samsung Frame Pro is a high-end smart TV that offers a range of upgrades and features, including a Wireless One Connect Box and a faster 144Hz maximum refresh rate. While the pricing may seem steep, it’s not outlandish considering the premium features and aesthetics. If you’re in the market for a high-end TV with a unique look and feel, the Samsung Frame Pro may be worth considering.

FAQs

Q: What sizes will the Samsung Frame Pro be available in?
A: The Frame Pro will be available in three sizes: 65-inch ($2,199), 75-inch ($3,199), and 85-inch ($4,299).

Q: What is the maximum refresh rate of the Samsung Frame Pro?
A: The maximum refresh rate of the Samsung Frame Pro is 144Hz.

Q: What is the difference between The Frame Pro and the standard 2025 Frame?
A: The Frame Pro has a faster 144Hz maximum refresh rate and a Wireless One Connect Box, which allows for a cable-free setup.

Q: Is the Samsung Frame Pro available to order now?
A: Yes, the Frame Pro is available to order now, with a larger 85-inch model coming "soon" according to Samsung.

Q: Does the Samsung Frame Pro have a matte, anti-glare display?
A: Yes, the Frame Pro has a matte, anti-glare display that helps the artwork it’s showcasing look more realistic and less like a lit-up screen.

LLMs Outperform Same-Size Models with IDA

Deep Cogito Releases Novel Open Large Language Models with General Superintelligence Claims

Deep Cogito has released several open large language models (LLMs) that outperform competitors and claim to represent a step towards achieving general superintelligence.

Overview of Deep Cogito’s Open LLMs

The San Francisco-based company, which states its mission is “building general superintelligence,” has launched preview versions of LLMs in 3B, 8B, 14B, 32B, and 70B parameter sizes. Deep Cogito asserts that “each model outperforms the best available open models of the same size, including counterparts from LLAMA, DeepSeek, and Qwen, across most standard benchmarks.”

Iterated Distillation and Amplification (IDA)

Central to this release is a novel training methodology called Iterated Distillation and Amplification (IDA). Deep Cogito describes IDA as “a scalable and efficient alignment strategy for general superintelligence using iterative self-improvement”. This technique aims to overcome the inherent limitations of current LLM training paradigms, where model intelligence is often capped by the capabilities of larger “overseer” models or human curators.

The IDA process involves two key steps iterated repeatedly:

  • Amplification: Using more computation to enable the model to derive better solutions or capabilities, akin to advanced reasoning techniques.
  • Distillation: Internalising these amplified capabilities back into the model’s parameters.

Deep Cogito says this creates a “positive feedback loop” where model intelligence scales more directly with computational resources and the efficiency of the IDA process, rather than being strictly bounded by overseer intelligence.

Capabilities and Performance of Deep Cogito Models

The newly released Cogito models – based on Llama and Qwen checkpoints – are optimised for coding, function calling, and agentic use cases. A key feature is their dual functionality: “Each model can answer directly (standard LLM), or self-reflect before answering (like reasoning models),” similar to capabilities seen in models like Claude 3.5. However, Deep Cogito notes they “have not optimised for very long reasoning chains,” citing user preference for faster answers and the efficiency of distilling shorter chains.

Benchmark Results

Extensive benchmark results are provided, comparing Cogito models against size-equivalent state-of-the-art open models in both direct (standard) and reasoning modes. Across various benchmarks (MMLU, MMLU-Pro, ARC, GSM8K, MATH, etc.) and model sizes (3B, 8B, 14B, 32B, 70B,) the Cogito models generally show significant performance gains over counterparts like Llama 3.1/3.2/3.3 and Qwen 2.5, particularly in reasoning mode.

Conclusion

This release is labelled a preview, with Deep Cogito stating they are “still in the early stages of this scaling curve”. They plan to release improved checkpoints for the current sizes and introduce larger MoE models (109B, 400B, 671B) “in the coming weeks / months”. All future models will also be open-source.

FAQs

Q: What is Iterated Distillation and Amplification (IDA)?
A: IDA is a novel training methodology that uses iterative self-improvement to overcome the limitations of current LLM training paradigms.

Q: What are the benefits of IDA?
A: IDA creates a “positive feedback loop” where model intelligence scales more directly with computational resources and the efficiency of the IDA process.

Q: What are the Cogito models optimised for?
A: The Cogito models are optimised for coding, function calling, and agentic use cases.

Q: What are the benchmark results?
A: The Cogito models generally show significant performance gains over counterparts like Llama 3.1/3.2/3.3 and Qwen 2.5, particularly in reasoning mode.

Q: What is the next step for Deep Cogito?
A: Deep Cogito plans to release improved checkpoints for the current sizes and introduce larger MoE models (109B, 400B, 671B) “in the coming weeks / months”.

Samsung Adds Google’s Gemini to Ballie

0

Samsung Partners with Google to Enhance Home Robot Ballie with Gemini AI

Introduction to Ballie and Gemini AI

Samsung is integrating Google’s Gemini AI into its home robot Ballie through a partnership with Google Cloud. This collaboration will enable users to ask Ballie various questions and receive answers from Gemini.

Multimodal Capabilities and Integration with Ballie

Samsung aims to tap into Gemini’s multimodal capabilities to enhance Ballie’s abilities. The company will pair its own AI with Google’s to enable audio and video inputs for Ballie to answer different questions. For instance, users will be able to ask the bot, "How am I looking?" and it will provide outfit suggestions using its camera and visual intelligence.

Examples of Gemini’s Capabilities on Ballie

  • Ballie will be able to provide health-related recommendations, such as exercise suggestions and ways to improve sleep.
  • Users will be able to ask the robot a range of general knowledge queries.

Benefits of the Partnership

As stated by Yongjae Kim, EVP of Samsung’s visual display business, "Through this partnership, Samsung and Google Cloud are redefining the role of AI in the home. By pairing Gemini’s powerful multimodal reasoning with Samsung’s AI capabilities in Ballie, we’re leveraging the power of open collaboration to unlock a new era of personalized AI companion — one that moves with users, anticipates their needs, and interacts in more dynamic and meaningful ways than ever before."

Availability and Future Plans

Samsung has been showcasing different versions of Ballie at trade shows like CES for years. The company has announced that Ballie will finally reach consumers in South Korea and the U.S. in the first half of this year. Samsung has also partnered with Google to integrate Gemini with its Galaxy series smartphones, starting with the Galaxy S24. Additionally, the two companies are working on an XR device, and Gemini may be central to that experience.

Conclusion

The partnership between Samsung and Google will bring significant enhancements to Ballie, enabling users to interact with the robot in more dynamic and meaningful ways. The integration of Gemini’s multimodal capabilities will provide users with a more personalized and intuitive experience.

Frequently Asked Questions

Q: What is Gemini AI?

A: Gemini is a multimodal AI developed by Google that can understand and respond to various inputs, including audio and video.

Q: What is Ballie?

A: Ballie is a home robot developed by Samsung that can perform various tasks and interact with users.

Q: What are the benefits of the partnership between Samsung and Google?

A: The partnership will bring significant enhancements to Ballie, enabling users to interact with the robot in more dynamic and meaningful ways. The integration of Gemini’s multimodal capabilities will provide users with a more personalized and intuitive experience.

Q: When will Ballie be available to consumers?

A: Ballie will be available to consumers in South Korea and the U.S. in the first half of this year.

Q: What other devices will Gemini be integrated with?

A: Gemini will be integrated with Samsung’s Galaxy series smartphones, starting with the Galaxy S24, and an XR device.

NVIDIA Brings Agentic AI Reasoning to Enterprises

AI Agents Driving New Enterprise Applications

NVIDIA is collaborating with Google Cloud to bring agentic AI to enterprises seeking to locally harness the Google Gemini family of AI models using the NVIDIA Blackwell HGX and DGX platforms and NVIDIA Confidential Computing for data safety.

Confidential Computing for Data Safety

With the NVIDIA Blackwell platform on Google Distributed Cloud, on-premises data centers can stay aligned with regulatory requirements and data sovereignty laws by locking down access to sensitive information, such as patient records, financial transactions, and classified government information. NVIDIA Confidential Computing also secures sensitive code in the Gemini models from unauthorized access and data leaks.

Unlocking the Full Potential of Agentic AI

“By bringing our Gemini models on premises with NVIDIA Blackwell’s breakthrough performance and confidential computing capabilities, we’re enabling enterprises to unlock the full potential of agentic AI,” said Sachin Gupta, vice president and general manager of infrastructure and solutions at Google Cloud. “This collaboration helps ensure customers can innovate securely without compromising on performance or operational ease.”

AI Agents Driving New Enterprise Applications

Agentic AI systems can reason, adapt, and make decisions in dynamic environments. For example, in enterprise IT support, an agentic AI system can diagnose issues, execute fixes, and escalate complex problems autonomously. Similarly, in finance, an agentic AI system can investigate anomalies and take proactive measures such as blocking transactions or adjusting fraud detection rules in real-time.

The On-Premises Dilemma

While many can already use the models with multimodal reasoning, those with stringent security or data sovereignty requirements have yet been unable to do so. With this announcement, Google Cloud will be one of the first cloud service providers to offer confidential computing capabilities to secure agentic AI workloads across every environment – whether cloud or hybrid.

AI Observability and Security for Agentic AI

Scaling agentic AI in production requires robust observability and security to ensure reliable performance and compliance. Google Cloud today announced a new GKE Inference Gateway built to optimize the deployment of AI inference workloads with advanced routing and scalability.

Conclusion

This collaboration between NVIDIA and Google Cloud brings agentic AI to enterprises seeking to locally harness the Google Gemini family of AI models. With confidential computing capabilities, enterprises can unlock the full potential of agentic AI while maintaining data safety and security.

FAQs

Q: What is agentic AI?
A: Agentic AI systems can reason, adapt, and make decisions in dynamic environments, unlike traditional AI models that perceive or generate based on learned knowledge.

Q: What is confidential computing?
A: Confidential computing ensures that sensitive code in AI models and data remain secure and cannot be viewed or modified.

Q: What are the benefits of this collaboration?
A: This collaboration enables enterprises to innovate securely without compromising on performance or operational ease, while maintaining data safety and security.

Q: What is the GKE Inference Gateway?
A: The GKE Inference Gateway is a new solution that optimizes the deployment of AI inference workloads with advanced routing and scalability.