From Digital Age to Nano Age. WorldWide.

Tag: engine

Robotic Automations

Apple teased AI improvements, including the M4's neural engine, at its iPad event | TechCrunch


Apple isn’t yet ready to unveil its broader AI strategy — it’s saving that for its Worldwide Developer Conference in June — but the tech giant did make sure to mention AI technologies across its device lineup at its iPad event on Tuesday. The company touted a new iPad Air as “an incredibly powerful device for AI;” its AI-powered features like visual lookup, subject lift and live text capture, among others; and, of course, its upgraded M4 chip, which features a neural engine that’s “dedicated to the acceleration of AI workloads.”

Tuesday’s theme focused on the new hardware devices themselves — devices like new iPads and an updated Apple Pencil — not on the AI advances Apple is making under the hood. But Apple made sure to note anytime its hardware advances could also help power AI technologies.

For instance, the iPad Air’s update, which now includes the M2 with a faster CPU, GPU and neural engine, was described as offering “powerful machine learning features,” like visual lookup, which can identify objects in photos; an AI-powered tool that can lift out a photo’s subject; and live text capture, which can copy, share, look up and translate text within the camera frame.

When Apple didn’t have its own AI tech to point to, it referenced third parties. While talking about the iPad Air, for example, Apple gave a shout-out to Pixelmator’s Photomator, which uses AI models trained on over 20 million professional images to improve photos with a single click.

Meanwhile, the iPad Pro jumped from being powered by the M2 to the new M4, the latest generation of Apple silicone, with a new CPU, next-gen GPU, and next-generation ML accelerators that Apple claims will deliver up to 50% faster performance than the M2. Of course, the company also played up the chip’s neural engine, or NPU, which is “dedicated to the acceleration of AI workloads.”

“Now while the chip industry is just starting to add NPUs to some of their processors, we’ve been including our industry-leading neural engine in our chip for years,” said John Ternus,  Apple SVP, Hardware Engineering, during the event.

Consumers, however, are waiting to see what sort of use cases are in store for these hardware advances — and that’s something Apple didn’t yet go into detail about, despite having the opportunity to tease an iPadOS release with new AI features or other developer-focused announcements.

Instead, Apple ran through iPadOS’ existing features, like multitasking view Stage Manager and a display mode for creatives, dubbed Reference Mode.

Still, the company hinted that improved AI capabilities would soon be in the hands of iPadOS app developers, noting that the operating system software offers advanced frameworks, like CoreML, and that developers would be able to tap into its neural engine to deliver “powerful AI features right on device.”

In other apps, like Logic Pro, Apple added new AI-powered session players like a bass and keyboard player who can join a performance alongside the drummer already available. These AI-driven backing band members can also respond to feedback provided in the app, Apple said. The company mentioned, too, a machine learning-backed Logic Pro plug-in, ChromaGlow, for adding warmth to tracks.

Image Credits: Apple

Apple made note of how AI could solve problems in areas like photography, like when trying to scan documents using a device’s camera.

“We’ve all had the experience of trying to scan a document in certain lighting conditions where it’s hard to avoid casting a shadow,” Ternus said. “The new Pro solves this problem. It uses AI to automatically detect documents like forms and receipts. If shadows are in the way it instantly takes multiple photos with the new adaptive flash. The frames are stitched together and the result is a dramatically better scan.”

Though none of the mentions of AI stood out as being over-the-top breakthroughs, they suggested that Apple’s style would be to note AI improvements as it related to upgrading the consumer experience of using its devices.

We expect to hear a lot more about Apple’s AI plans at WWDC, where the company may even announce an AI-powered Siri or partnership with an AI provider like Google or OpenAI, rumors suggest.

 


Software Development in Sri Lanka

Robotic Automations

Danti's natural language search engine for Earth data soars with $5M in new funding | TechCrunch


Danti, an artificial intelligence company building a superpower search engine for Earth data, has brought on prominent defensive tech investor Shield Capital as it looks to scale its technology for government customers.

Founded by Jesse Kallman in early 2023, Danti has developed a natural language search engine for data that has historically been highly siloed, like satellite imagery, collating it with other commercial and government sources to report back across multiple sources and domains.

For example, an analyst can pose a complex question in simple language, like “What are the latest tank movements in Eastern Ukraine?” and receive in turn straightforward answers collated across data sources.

The idea is to empower a single analyst to do more, Kallman said in a recent interview. While American adversaries are throwing manpower at the problem of analyzing huge amounts of data, Danti aims to help “one analyst do the work of 10 or 15,” he said. It means that a relatively straightforward question — where is a particular ship off the coast of Lagos, Nigeria, for example — can potentially be answered in seconds, rather than hours.

“We’re not replacing the analysts,” Kallman clarified. “We’re helping them do their work way faster, so that they can get to the part that humans are way better at, which is synthesizing and deciding, ‘What do I now do about this information? How do I want to report on it?’ “

Among the startup’s early customers is the U.S. Space Force, which is using Danti’s product to help officers easily search and share data. The use of natural language models in the search engine means an intuitive, straightforward user experience; no doubt this is paramount in high pressure situations where analysts must make complex decisions, but have little time to trawl through reams of satellite or drone data.

Right now, Danti is squarely focused on government, though in the longer term it plans to roll out a version of its product for commercial industry. This version would focus on property records, parcel information, and risk data, to serve markets like electric utilities and insurance, Kallman said. Customers will also be able to connect their own information into Danti’s engine to use its natural language processing to query their own data.

The $5 million round was led by Shield Capital and includes participation from the startup’s existing investors Tech Square Ventures, Humba Ventures and Leo Polovets, Space.VC, and Radius Capital. Kallman said the startup looked deliberately for a defense-focused fund to lead their next round, particularly as the company looks to execute its government go-to-market plan and scale its engineering team.

Since last summer, when the company announced its $2.75 million pre-seed, the team has grown to over twenty people, and Kallman said the engineering team will grow even more with the new injection of funds.


Software Development in Sri Lanka

Back
WhatsApp
Messenger
Viber