"Today's AIs are book smart," reports the Wall Street Journal. "Everything they know they learned from available language, images and videos. To evolve further, they have to get street smart." And that requires "world models," which are "gaining momentum in frontier research and could allow technology to take on new roles in our lives." The key is enabling AI to learn from their environments and faithfully represent an abstract version of them in their "heads," the way humans and animals do. To do it, developers need to train AIs by using simulations of the world. Think of it like learning to drive by playing "Gran Turismo" or learning to fly from "Microsoft Flight Simulator." These world models include all the things required to plan, take actions and make predictions about the future, including physics and time... There's an almost unanimous belief among AI pioneers that world models are crucial to creating next-generation AI. And many say they will be critical to someday creating better-than-human "artificial general intelligence," or AGI. Stanford University professor and AI "godmother" Fei-Fei Li has raised $230 million to launch world-model startup World Labs... Google DeepMind researchers set out to create a system that could generate real-world simulations with an unprecedented level of fidelity. The result, Genie 3 — which is still in research preview and not publicly available — can generate photo-realistic, open-world virtual landscapes from nothing more than a text prompt. You can think of Genie 3 as a way to quickly generate what's essentially an open-world videogame that can be as faithful to the real world as you like. It's a virtual space in which a baby AI can endlessly play, make mistakes and learn what it needs to do to achieve its goals, just as a baby animal or human does in the real world. That experimentation process is called reinforcement learning. Genie 3 is part of a system that could help train the AI that someday pilots robots, self-driving cars and other "embodied" AIs, says project co-lead Jack Parker-Holder. And the environments could be filled with people and obstacles: An AI could learn how to interact with humans by observing them moving around in that virtual space, he adds. "It isn't clear whether all these bets will lead to the superintelligence that corporate leaders predict," the article concedes. "But in the short term, world models could make AIs better at tasks at which they currently falter, especially in spatial reasoning."

Read more of this story at Slashdot.

The BBC reports that a million-year-old human skull found in China suggests that the human species "began to emerge at least half a million years earlier than we thought, researchers are claiming in a new study." It also shows that we co-existed with other sister species, including Neanderthals, for much longer than we've come to believe, they say. The scientists claim their analysis "totally changes" our understanding of human evolution and, if correct, it would certainly rewrite a key early chapter in our history. But other experts in a field where disagreement over our emergence on the planet is rife, say that the new study's conclusions are plausible but far from certain. The discovery, published in the leading scientific journal Science, shocked the research team, which included scientists from a university in China and the UK's Natural History Museum. "From the very beginning, when we got the result, we thought it was unbelievable. How could that be so deep into the past?" said Prof Xijun Ni of Fudan University, who co-led the analysis. "But we tested it again and again to test all the models, use all the methods, and we are now confident about the result, and we're actually very excited." Thanks to long-time Slashdot reader sinij for sharing the article.

Read more of this story at Slashdot.

Six months ago California had 48% more public and "shared" private EV chargers than gasoline nozzles. (In March California had 178,000 public and shared private EV chargers, versus about 120,000 gas nozzles.) Since then they've added 23,000 more public/shared charging ports — and announced this week that there's now 68% more EV charger ports than the number of gasoline nozzles statewide. "Thanks to the state's ever-expanding charger network, 94% of Californians live within 10 minutes of an EV charger," according to the announcement from the state's energy policy agency. And the California Energy Commission staff told CleanTechnica they expect more chargers in the future. "We are watching increased private investment by consortiums like IONNA and OEMs like Rivian, Ford, and others that are actively installing EV charging stations throughout the state." Clean Technica notes in 2019, the state had roughly 42,000 charging ports and now there are a little over 200,000. (And today there's about 800,000 home EV chargers.) This week California announced another milestone: that in 2024 nearly 23% of all the state's new truck sales — that's trucks, buses, and vans — were zero-emission vehicles. (The state subsidizes electric trucks — $200 million was requested on the program's first day.) Greenhouse gas emissions in California are down 20% since 2000 — even as the state's GDP increased 78% in that same time period all while becoming the world's fourth largest economy. The state also continues to set clean energy records. California was powered by two-thirds clean energy in 2023, the latest year for which data is available — the largest economy in the world to achieve this level of clean energy. The state has run on 100% clean electricity for some part of the day almost every day this year. "Last year, California ran on 100% clean electricity for the equivalent of 51 days," notes another announcement, which points out California has 15,763 MW of battery storage capacity — roughly a third of the amount projected to be needed by 2045.

Read more of this story at Slashdot.

Paris-based AI giant Mistral "is pushing to improve its models," reports the Wall Street Journal, "by looking inside legacy enterprises that hold some of the world's last untapped data reserves...." Mistral's approach will be to form partnerships with enterprises to further train existing models on their own proprietary data, a phenomenon known as post-training... [At Dutch chip-equipment company ASML], Mistral embeds its own solutions architects, applied AI engineers and applied scientists into the enterprise to work on improving models with the company's data. [While Mistral sells some models under a commercial license], this co-creation strategy allows Mistral to make money off the services side of its business and afford to give away its open source AI free of charge, while improving model performance for the customer with more industry context... This kind of hand-holding approach is necessary for most companies to tackle AI successfully, said Arthur Mensch [co-founder and chief executive of Mistral]. "The very high-tech companies [and] a couple of banks are able to do it on their own. But when it comes to getting some [return on investment] from use cases, in general, they fail," he said. Mensch attributes that in part to a mismatch between expectations and reality. "The curse of AI is that it looks like magic. So you can very quickly make something that looks amazing to your boss," but it doesn't scale or work more broadly, he said. In other cases, enterprises simply might not know what to focus on. For example, it is a mistake to think equipping all employees with a chatbot will create meaningful gains on the bottom line, he said. Mensch said to fully take advantage of AI, companies will have to rethink organizational structures. With information flowing more easily, they could require fewer middle managers, for example. There is a lot of work yet to do, Mensch said, but in a large sense, the future of AI development now lies inside the enterprise itself. "This is a pattern that we've seen with many of our customers: At some point, the capabilities of the frontier model can only be increased if we partner," he said.

Read more of this story at Slashdot.

Long-time Slashdot reader theodp says America's Patent and Trademark Office (USPTO) has granted a patent to Tableau (Salesforce's visual analytics platform) — for a patent covering "Data Processing For Visualizing Hierarchical Data": "A provided data model may include a tree specification that declares parent-child relationships between objects in the data model. In response to a query associated with objects in the data model: employing the parent-child relationships to determine a tree that includes parent objects and child objects from the objects based on the parent-child relationships; determining a root object based on the query and the tree; traversing the tree from the root object to visit the child objects in the tree; determining partial results based on characteristics of the visited child objects such that the partial results are stored in an intermediate table; and providing a response to the query that includes values based on the intermediate table and the partial results." A set of 15 simple drawings is provided to support the legal and tech gobbledygook of the invention claims. A person can have a manager, Tableau explains in Figures 5-6 of its accompanying drawings, and that manager can also manage and be managed by other people. Not only that, Tableau illustrates in Figures 7-10 that computers can be used to count how many people report to a manager. How does this magic work, you ask? Well, you "generate [a] tree" [Fig. 13] and "traverse a tree" [Fig. 15], Tableau explains. But wait, there's more — you can also display the people who report to a manager in multi-level or nested pie charts (aka Sunburst charts), Tableau demonstrates in Fig. 11. Interestingly, Tableau released a "pre-Beta" Sunburst chart type in late April 2023 but yanked it at the end of June 2023 (others have long-supported Sunburst charts, including Plotly). So, do you think Tableau should be awarded a patent in 2025 on a concept that has roots in circa-1921 Sunburst charts and tree algorithms taught to first-year CS students in circa-1975 Data Structures courses?

Read more of this story at Slashdot.

"We've decided to support image-based search," announced the product manager for Firefox Search. Powered by the AI-driven Google Lens search technology, they promise the new feature offers "a frictionless, fast, and a curiosity-sparking way to (as Google puts it) 'search what you see'." With just a right-click on any image, you'll be able to: - Find similar products, places, or objects - Copy, translate, or search text from images - Get inspiration for learning, travel, or shopping Look for the new "Search Image with Google Lens" option in your right-click menu (tagged with a NEW badge at first). This is a desktop-only feature, and it will start gradually rolling out worldwide. Note: Google must be set as your default search engine for this feature to appear. We'll be listening closely to your feedback as we roll it out. Some of the things we're wondering about: Does the placement in the context menu align with your expectations? Would you prefer the option to choose your visual search provider? Where else would you like entry points to visual search (e.g. when you open a new tab, in the address bar, on mobile devices, etc.) We can't wait to hear your thoughts as the rollout begins! Some thoughts from WebProNews: Mozilla emphasizes that this is an opt-in feature, giving users control over activation, which aligns with the company's longstanding commitment to privacy and user agency. Yet, for industry observers, this partnership with Google raises intriguing questions about competitive dynamics in the browser space, where Firefox has historically positioned itself as an independent alternative to Chrome... This move comes at a time when browsers are increasingly becoming platforms for AI-driven enhancements, as evidenced by recent updates in competitors like Microsoft's Edge, which integrates Copilot AI. Mozilla's decision to leverage Google Lens rather than developing an in-house solution could be seen as a pragmatic step to accelerate feature parity, especially given Firefox's smaller market share. Insiders note that by tapping into established technologies, Mozilla can focus resources on core strengths like privacy protections, potentially attracting users disillusioned with data-heavy ecosystems... While mobile users might feel left out, the phased rollout over the next few weeks allows for feedback loops through community channels, a hallmark of Mozilla's open-source ethos. Data from similar integrations in other browsers suggests visual search can boost engagement by 15-20%, per industry reports, though Mozilla has not disclosed specific metrics yet... Looking ahead, Mozilla's strategy appears geared toward incremental innovations that bolster user retention without alienating its privacy-focused base. If successful, this could help Firefox claw back some ground against Chrome's dominance, estimated at over 60% market share. For now, the feature's gradual deployment invites ongoing dialogue, underscoring Mozilla's community-driven model in an industry often criticized for top-down decisions.

Read more of this story at Slashdot.

Cristian Fleming paid around $70,000 for one of Fisker Ocean's electric mid-size crossover SUVs. Seven months later the company filed for bankruptcy in June of 2024, reports the Verge, "having only delivered 11,000 vehicles." "Early adopters were left with cars plagued by battery failures, glitchy software, inconsistent key fobs, and door handles that did not always open. With the company gone, there was no way to fix any issues." Regulators logged dozens of complaints as replacement parts vanished. Passionate owners who spent top dollar on high-end trims saw their cars reduced to expensive driveway ornaments. Rather than accept defeat, thousands of Ocean owners have organized into their own makeshift car company. The Fisker Owners Association (FOA) is a nonprofit that's launched third-party apps, built a global parts supply chain, and came together around a future for their orphaned vehicles. It's part car club, part tech startup, part survival mission. Fleming now serves as the organization's president... FOA calls itself the first entirely owner-controlled EV fleet in history. So far, 4,055 Ocean owners have signed up, paying $550 a year in dues that the group estimates will raise around $3 million annually, about 0.1 percent of Fisker's peak valuation. Only verified Ocean owners can become full members, but anyone can donate. The grassroots effort has precedent — DeLorean diehards and Saab enthusiasts have kept their favorite brands alive after factory closures. But those efforts focused on preserving aging vehicles. FOA is attempting something different: real-time software updates and hardware improvements for a connected, two-year-old EV fleet... The organization has spawned three separate companies. Tsunami Automotive handles parts in North America while Tidal Wave covers Europe, scavenging insurance auctions and contracting with tooling manufacturers to reproduce components. UnderCurrent Automotive, run by former Google and Apple engineers, focuses on software solutions. UnderCurrent's first product is OceanLink Pro, a third-party mobile app now used by over 1,200 members that restores basic EV features, such as remote battery monitoring and climate control. A companion device called OceanLink Pulse adds wireless CarPlay and Android Auto, with plans for future upgrades including keyless entry. "Those are things you would have expected to be in a $70,000 luxury car," says Clint Bagley [FOA's treasurer]. "But, you know, we're happy to provide what the billion-dollar automaker apparently couldn't."

Read more of this story at Slashdot.