Tag: IFTTT

Tucked away in the Swedish countryside is a facility quietly reshaping the future of global mobility. Owned by the Research Institutes of Sweden (RISE), AstaZero has just unveiled the world’s most advanced connected vehicle proving ground—an ambitious leap into a 6G-powered future where every movement on the road could be coordinated, controlled, and optimized in real time.

AstaZero is not an average vehicle test track. It is a full-scale, independent research environment built to test the automated transport systems of tomorrow to ensure confidence and safety. Think of it as a real-world lab where self-driving cars, AI-powered drones, and connected emergency vehicles are pushed to their limits.

At the heart of this latest breakthrough are multiple 5G networks and a cutting-edge computing facility—marking a first for any open, brand-neutral proving ground. It enables split-second decision-making and ultra-reliable connectivity between vehicles, emergency teams, pedestrians, infrastructure, and traffic systems.

That matters more than ever. With 3G networks being phased out globally, mission-critical systems like ambulances, fire trucks, and police vehicles are under pressure to modernize. AstaZero’s newly launched facility provides the first real opportunity to test innovative systems in controlled yet dynamic, real-life scenarios.

AstaZero’s new infrastructure is not just about faster speeds—it is about smarter, safer reactions. Powered by edge computing, vehicles can now process data locally instead of relying on far-off cloud centers. That means a self-driving car can respond instantly to a pedestrian stepping into the street or adjust to a new traffic signal before the driver sees it.

Without advanced, integrated testing, safer roads remain a dream. CEO of RISE AstaZero Peter Janevik explained the implications of this breakthrough, telling Gizmodo, “In the future, communication might not always originate from the sensors on the vehicle itself, but instead from sensors mounted on connected infrastructure or from the sensors of another vehicle. In these types of systems, three key factors are crucial: reliability, ultra-fast communication, and intelligent decision-making.”

In June, AstaZero said it had reached 99.999% system reliability in connected vehicle communication, a first for the industry. That is the level of consistency required for “mission-critical” scenarios, where even a split-second failure could cost lives.

When asked what type of real-world scenarios are most challenging to simulate at AstaZero and how they overcome them, Janevik described the complexity of multiple testing domains with a future scenario:

An automated drone providing safety surveillance is deployed over an accident scene by a rescue crew upon arrival. The footage is used by both the rescue crew to assess and follow the situation, but also by central management, which needs to make decisions on things such as rerouting of traffic and the deployment of further teams and other authorities like police and medical teams. Then imagine that the drone also creates a local map update with static objects such as a crashed vehicle or cones for traffic redirection and dynamic ones such as personnel or fires. Imagine that this map is also used for warnings and rerouting of automated as well as manually driven vehicles.

Heads-up displays may be the latest step in this direction, with emergency information scrolling along the lower edge of the windshield and not on overhead traffic signs or infotainment screens. To ensure such a complex system works, the testing and design teams need to factor in elements like connectivity disruption and technology integration across numerous manufacturers and telecom companies, which is what AstaZero offers.

Beyond roads and intersections, AstaZero’s proving ground is designed to test limitless scenarios. Whether cyclists swerving through traffic or simulated pedestrians crossing at unpredictable times, the site can orchestrate complex environments. Janevik says, “We test collision avoidance technology to auto-brake vehicles for different scenarios, but more importantly, the site provides robust testing to ensure highly repeatable results in a wider spectrum of conditions.”

By using AI, drones, and robotic systems—like digital twins and virtual modeling—for advanced scenario computations and simulations, the site assists engineers in pursuing advances in chip manufacturing, so designs keep track with forthcoming technologies. Janevik believes in the impact of this approach on “unique testing scenarios for smaller machine learning models with AI-based decision-making to prove that these can make the right decisions with ongoing updates.”

The RISE facility’s goal is to test components in a hardware loop in the vehicle in real-world scenarios. Testing also accounts for degraded conditions—such as lost connectivity—to prepare for actual challenges. The only limits are what the engineers can imagine, and Janevik sees this as their goal—to live their vision and help societies accelerate into safe, sustainable, and automated transportation systems of the future.

This is especially critical in Europe, where road fatality statistics have stagnated. While there was a 10% drop in EU road deaths between 2019 and 2023, the latest figures show only a 1% decrease. With 83% of fatal pedestrian accidents occurring in urban areas and a stubborn plateau in progress, new solutions are needed. As EU Commissioner for Sustainable Transport and Tourism Apostolos Tzitzikostas has said, “Too many lives are still lost on our roads every year.”

AstaZero stands out for being brand-agnostic. Any vehicle manufacturer, telecom provider, or AI developer can pay to use the facility to test and refine their systems. That neutral status is intended to ensure consistency and fairness across global standards, which is especially important as the European New Car Assessment Programme rolls out new vehicle-to-everything benchmarks between 2026 and 2032. Already a recognized test organization by the Global Certification Forum, AstaZero has taken a lead role in helping shape those standards.

The AstaZero proving ground does not just test how cars perform—it tests how they think, communicate, and collaborate. With edge computing enabling decentralized, real-time responses, the next generation of smart vehicles will be able to prevent accidents before they happen, minimize traffic delays, and drastically improve energy efficiency.

The transformation underway in finance should give pause to anyone who still thinks artificial intelligence is a distant threat. On July 15, Wall Street met its most overqualified and tireless intern. AI safety and research company Anthropic, a chief rival to OpenAI, unveiled its new “Financial Analysis Solution,” an enhanced version of its Claude AI assistant designed to take over the research, modeling, and compliance grunt work that finance teams typically rely on junior analysts to perform.

This specialized version of Claude can now parse corporate earnings calls, scan vast financial data warehouses, run complex Monte Carlo simulations (a sophisticated technique that plays out a financial “what if” game thousands of times to map all possibilities), and produce investment memos that look like they came from a human who has not slept in three days. The human in question, however, may not be needed much longer.

The announcement came with powerful testimonials from industry giants. Bridgewater Associates, one of the largest and most influential hedge funds in the world, is already a user.

“We’ve been developing capabilities powered by Claude since 2023,” said Aaron Linsky, CTO of AIA Labs at Bridgewater. “Claude powered the first versions of our Investment Analyst Assistant, which streamlined our analysts’ workflow by generating Python code, creating data visualizations, and iterating through complex financial analysis tasks with the precision of a junior analyst.”

The translation is clear: Claude is already doing the work of entry level employees at the world’s most elite firms.

Claude Is Now a Finance Analyst in a Box

Anthropic claims that its latest model, Claude 4, outperforms OpenAI’s GPT-4 and other rivals on specialized financial tasks. In one benchmark, Claude scored 83 percent accuracy on complex Excel modeling challenges that simulate real world investment cases. This means Claude can now perform tasks that are the bedrock of modern finance:

• Build and tweak the intricate financial models used to value companies or forecast cash flows.
• Analyze quarterly earnings calls and summarize key takeaways instantly.
• Pull data from complex data warehouses like Snowflake or Databricks (think of these as giant, centralized digital libraries for a company’s financial information) and visualize it on demand.
• Draft institutional quality pitch decks and investment memos.
• Write Python code, a popular programming language, to automate tedious number crunching tasks.

Claude does not just assist humans; it executes entire workflows. That’s why Anthropic is positioning it less as a simple chatbot and more as an enterprise grade workhorse.

213,000 Hours Gone and Not Coming Back

The statistics provided by early adopters are staggering. Norway’s sovereign wealth fund, NBIM, which manages one of the largest investment funds globally, says Claude has already replaced the equivalent of over 213,000 work hours across its finance and risk teams. CEO Nicolai Tangen quantified the productivity gains at approximately 20 percent and added that Claude now automates the monitoring of news and earnings calls for 9,000 companies.

Meanwhile, insurance giant AIG says it is using Claude to transform its underwriting process. According to CEO Peter Zaffino, “We have been able to compress the timeline to review business by more than 5x (…) while simultaneously improving our data accuracy from 75% to over 90%.”

In the high stakes world of finance, speed and precision are everything. Claude appears to offer both, without demanding a bonus or taking a vacation.

The Death of the Analyst Track?

For decades, aspiring financiers have cut their teeth as entry level analysts at big banks and hedge funds. The work is notoriously brutal, defined by 80 hour weeks, endless Excel modeling, and all nighters building pitch decks that no one might read. But it has always been the essential rite of passage into a lucrative career.

Now, Claude does all of that. It does it faster, without typos, and without needing to impress a managing director.

While Anthropic insists the goal is to free up humans, it is clear where this is heading. Claude does not just save time on grunt work. It has the potential to replace the very need for junior analysts to be doing this work in the first place.

Anthropic is pitching this as a win-win. Companies save money and analysts get to “focus on higher level tasks.” But in a cutthroat industry where cost cutting is constant, those higher level tasks may not materialize fast enough to absorb a workforce whose primary function has been automated.

AI Isn’t Just Coming for Blue Collar Jobs Anymore

The finance industry has long assumed that its white collar ranks were safe from AI disruption. Automation might hit the factory floor or call centers, but not the corner office.

Claude directly challenges that assumption. And it is not alone. OpenAI is working with PwC on similar initiatives. Google is embedding its Gemini models into trading platforms. The race is on to see which AI company can reshape Wall Street first and reap the rewards. Claude’s edge might be its deep integration strategy. It connects with essential financial data sources from S&P Global and Morningstar and platforms like Palantir and Snowflake. Claude can even write compliance policies, thanks to partnerships with accounting firms like PwC and Deloitte.

In short, Claude is not just intelligent. It is deeply wired into the plumbing of modern finance.

A Future Where Your AI Writes the Memos

Anthropic said that Claude is now available on the AWS Marketplace, with Google Cloud availability coming soon. Companies can drop it into their research teams or build custom applications using Claude’s API, a tool that allows different software programs to communicate with each other. If a firm wants its AI to write underwriting policies or track complex ESG (Environmental, Social, and Governance) compliance, Claude can do that too.

“Claude provides the complete platform for financial AI,” the company said. “Every claim links directly to its original source for transparency, and complex analysis that normally takes hours happens in minutes.”

If that sounds like the future of finance, it probably is. But for thousands of young analysts hoping to climb the ladder, Claude may have just pulled the first few rungs out from under them.

If you want a job at McDonald’s today, there’s a good chance you’ll have to talk to Olivia. Olivia is not, in fact, a human being, but instead an AI chatbot that screens applicants, asks for their contact information and resumé, directs them to a personality test, and occasionally makes them “go insane” by repeatedly misunderstanding their most basic questions.

"You operate as an autonomous agent controlling a pursuit spacecraft."

This is the first prompt researchers used to see how well ChatGPT could pilot a spacecraft. To their amazement, the large language model (LLM) performed admirably, coming in second place in an autonomous spacecraft simulation competition.

Researchers have long been interested in developing autonomous systems for satellite control and spacecraft navigation. There are simply too many satellites for humans to manually control them in the future. And for deep-space exploration, the limitations of the speed of light mean we can’t directly control spacecraft in real time.

If we really want to expand in space, we have to let the robots make decisions for themselves.

To encourage innovation, in recent years aeronautics researchers have created the Kerbal Space Program Differential Game Challenge, a sort of playground based on the popular Kerbal Space Program video game to allow the community to design, experiment and test autonomous systems in a (somewhat) realistic environment. The challenge consists of several scenarios, like a mission to pursue and intercept a satellite and a mission to evade detection.

In a paper to be published in the Journal of Advances in Space Research, an international team of researchers described their contender: a commercially available LLM, like ChatGPT and Llama.

The researchers decided to use an LLM because traditional approaches to developing autonomous systems require many cycles of training, feedback and refinement. But the nature of the Kerbal challenge is to be as realistic as possible, which means missions that last just hours. This means it would be impractical to continually refine a model.

But LLMs are so powerful because they’re already trained on vast amounts of text from human writing, so in the best case scenario they need only a small amount of careful prompt engineering and a few tries to get the right context for a given situation.

But how can such a model actually pilot a spacecraft?

The researchers developed a method for translating the given state of the spacecraft and its goal in the form of text. Then, they passed it to the LLM and asked it for recommendations of how to orient and maneuver the spacecraft. The researchers then developed a translation layer that converted the LLM’s text-based output into a functional code that could operate the simulated vehicle.

With a small series of prompts and some fine-tuning, the researchers got ChatGPT to complete many of the tests in the challenge — and it ultimately placed second in a recent competition. (First place went to a model based on different equations, according to the paper).

And all of this was done before the release of ChatGPT’s latest model, version 4. There’s still a lot of work to be done, especially when it comes to avoiding "hallucinations" (unwanted, nonsensical output), which would be especially disastrous in a real-world scenario. But it does show the power that even off-the-shelf LLMs, after digesting vast amounts of human knowledge, can be put to work in unexpected ways.

This article was originally published in LiveScience. Read the original article here.

And all of this was done before the release of ChatGPT’s latest model, version 4. There’s still a lot of work to be done, especially when it comes to avoiding "hallucinations" (unwanted, nonsensical output), which would be especially disastrous in a real-world scenario. But it does show the power that even off-the-shelf LLMs, after digesting vast amounts of human knowledge, can be put to work in unexpected ways.

This article was originally published in LiveScience. Read the original article here.

Join our Space Forums to keep talking space on the latest missions, night sky and more! And if you have a news tip, correction or comment, let us know at: community@space.com.