Day: November 7, 2023

A machine-learning tool can easily spot when chemistry papers are written using the chatbot ChatGPT, according to a study published on 6 November in Cell Reports Physical Science. The specialized classifier, which outperformed two existing artificial intelligence (AI) detectors, could help academic publishers to identify papers created by AI text generators.

“Most of the field of text analysis wants a really general detector that will work on anything,” says co-author Heather Desaire, a chemist at the University of Kansas in Lawrence. But by making a tool that focuses on a particular type of paper, “we were really going after accuracy.”

The findings suggest that efforts to develop AI detectors could be boosted by tailoring software to specific types of writing, Desaire says. “If you can build something quickly and easily, then it’s not that hard to build something for different domains.”

The elements of style

Desaire and her colleagues first described their ChatGPT detector in June, when they applied it to Perspective articles from the journal Science. Using machine learning, the detector examines 20 features of writing style, including variation in sentence lengths, and the frequency of certain words and punctuation marks, to determine whether an academic scientist or ChatGPT wrote a piece of text. The findings show that “you could use a small set of features to get a high level of accuracy,” Desaire says.

In the latest study, the detector was trained on the introductory sections of papers from ten chemistry journals published by the American Chemical Society (ACS). The team chose the introduction because this section of a paper is fairly easy for ChatGPT to write if it has access to background literature, Desaire says. The researchers trained their tool on 100 published introductions to serve as human-written text, and then asked ChatGPT-3.5 to write 200 introductions in ACS journal style. For 100 of these, the tool was provided with the papers’ titles, and for the other 100, it was given their abstracts.

When tested on introductions written by people and those generated by AI from the same journals, the tool identified ChatGPT-3.5-written sections based on titles with 100% accuracy. For the ChatGPT-generated introductions based on abstracts, the accuracy was slightly lower, at 98%. The tool worked just as well with text written by ChatGPT-4, the latest version of the chatbot. By contrast, the AI detector ZeroGPT identified AI-written introductions with an accuracy of only about 35–65%, depending on the version of ChatGPT used and whether the introduction had been generated from the title or the abstract of the paper. A text-classifier tool produced by OpenAI, the maker of ChatGPT, also performed poorly — it was able to spot AI-written introductions with an accuracy of around 10–55%.

The new ChatGPT catcher even performed well with introductions from journals it wasn’t trained on, and it caught AI text that was created from a variety of prompts, including one aimed to confuse AI detectors. However, the system is highly specialized for scientific journal articles. When presented with real articles from university newspapers, it failed to recognize them as being written by humans.

Wider issues

What the authors are doing is “something fascinating,” says Debora Weber-Wulff, a computer scientist who studies academic plagiarism at the HTW Berlin University of Applied Sciences. Many existing tools try to determine authorship by searching for the predictive text patterns of AI-generated writing rather than by looking at features of writing style, she says. “I’d never thought of using stylometrics on ChatGPT.”

But Weber-Wulff points out that there are other issues driving the use of ChatGPT in academia. Many researchers are under pressure to quickly churn out papers, she notes, or they might not see the process of writing a paper as an important part of science. AI-detection tools will not address these issues, and should not be seen as “a magic software solution to a social problem.”

This article is reproduced with permission and was first published on January 27 2023.

• China revealed its bold plans to mass produce “advanced-level” humanoid robots by 2025. 
• China’s Ministry of Industry and Information Technology published a roadmap of its plans last week.
• The MIIT believes that humanoid robots will be as ”disruptive” as smartphones and electric vehicles.

China revealed ambitious plans to mass produce humanoid robots, which it believes will be as “disruptive” as smartphones. 

In an ambitious blueprint document published last week, China’s Ministry of Industry and Information Technology said the robots would “reshape the world.”

The MIIT believes that by 2025, the product would have reached “advanced level” and be “mass-produced,” the development goals listed in its roadmap said.

“They are expected to become disruptive products after computers, smartphones, and new energy vehicles,” a translation of the document added.

Per Bloomberg, the document was “short on details but big on ambition.” However, some Chinese companies are seemingly helping to tackle the country’s ambition in earnest.

For example, Chinese startup Fourier Intelligence said it would start mass producing its GR-1 humanoid robot by the end of this year, South China Morning Post reported. The Shanghai-based company told the publication it aspires to deliver thousands of robots in 2024 that can move at five kilometers an hour and carry 50 kilograms.

It’s not the only humanoid robot-maker that’s ramping up its efforts with the goal of mass production. US-based Agility Robotics is opening a robot factory in Oregon later this year, where it plans to build hundreds of its bipedal robots that can mimic human movements like walking, crouching, and carrying packages.

E-commerce giant Amazon is testing Agility Robotics’ Digit robot at a research and development center near Seattle to see how it can be used to automate its warehouses, but it’s only in the pilot phase. 

Agility Robotics CEO Damion Shelton told Insider: “In the near term, we expect a slow and steady uptick of Digit deployments.” He added: “We believe mass integration will eventually occur, but bipedal robots are still a relatively new advancement.” 

Even Tesla is developing its own humanoid robots called Optimus, or Tesla Bot, as Elon Musk revealed in 2021. However, it still has a long way to go before it’s ready for mass production as Musk said at a Tesla AI Day event in 2022 that it was the first time the prototype had walked “without any support” when it walked onto the stage. 

 

A bus charging wirelessly on a road equipped to charge electric vehicles in Balingen, Germany. (Electreon)

 

Wireless electric road charging has long been hailed as a potential holy grail for widespread electric vehicle adoption.

Later this year, the technology will get its first real test in the U.S., when Israeli company Electreon and the state of Michigan unveil a 1-mile stretch of road in Detroit that will allow drivers to charge on the go, marking the first public deployment in the country.

“You can go longer range,” said Stefan Tongur, vice president of business development at Electreon U.S. “You don’t need to have a huge battery, which drives costs down, and it makes it easier for the grid. At the end of the day, if we want to reach a high level of [EV] adoption, … charging is the main barrier.”

Electreon’s technology, known as dynamic charging, works similarly to the wireless pads that charge your smartphone. It embeds copper coils that are connected directly to the power grid into the pavement. Vehicles are equipped with receiver plates installed under the car or truck, so the charge passes through to the battery every time that vehicle drives over electrified roads.

It’s all operated remotely through Electreon’s software and servers. That means Electreon can dial down the charge when the power grid is stretched and dial up the power when there’s more capacity.

“As a vehicle has a small battery on board, it doesn’t really matter whether it charges in this mile or the next mile, but that could mean everything to the utility,” said Regan Zane, director of Utah State University’s Center for Advancing Sustainability through Powered Infrastructure for Roadway Electrification (ASPIRE), where Electreon has been testing its technology. “Having that flexible resource could be a game-changing solution for the utility.”

EV charging, range anxiety a critical factor

The deployment of dynamic charging in the U.S. comes as charging infrastructure and range anxiety continue to slow EV adoption in the country. 

While electric vehicle sales volume hit another record in the third quarter, accounting for 7.9% of total industry sales, according to Cox Automotive, drivers continue to express hesitation about the lack of chargers in place. 

In a recent Yahoo Finance-Ipsos poll, 77% of people surveyed said a lack of charging stations on the road or charging at home would discourage them from going electric.

Another factor is that roughly 30% of chargers aren’t working at any given moment, according to Akshay Singh, industrial and automotive principal at PwC.

“They’re not reliable — that’s a big, big factor,” said Singh. “The problem is that if you look at a lot of the [charging] companies, they don’t have a robust maintenance program in place.”

Singh added that high capital costs tied to electric vehicle chargers have also slowed deployment, with a 350-kilowatt charger costing “upwards” of $250,000. Charge point operators need 15% to 20% utilization just to break even, he said.

PwC estimates the EV charging market would need to grow five-fold to $40 billion in order to meet the demand that would come with 50% adoption in the US.

Cost hurdles continue for wireless EV charging

Dynamic charging technology has already been deployed in nearly half a dozen countries. Sweden has led the way, with public buses and trucks already charging on the go along a short stretch of highway on the island of Gotland. Electreon’s technology has also been deployed in Israel, Germany, and Italy. In October, the company announced it had signed a memorandum of understanding with the government of China’s Shandong Province to integrate Electreon’s technology in the world’s largest EV market.

Dynamic charging in Detroit is being deployed in partnership with Ford (F), and the project is scheduled to be completed by the end of the year, according to the Michigan Department of Transportation. 

Congress is also betting on the technology with a bill that proposes setting aside $250 million in grants for wireless EV charging initiatives.

While dynamic charging may finally be generating interest globally, the price to implement the technology — a cost of $1.2 million per mile — remains a hurdle.

Challenging economics contributed to Qualcomm Technologies (QCOM) selling its Halo wireless EV charging business back in 2019 as well as Bombardier (BDRBF) selling its EV charging systems business, along with its entire transportation business to Alstom in 2021.

Aside from Electreon, Renault and Swedish trucking giant Scania AB have continued with their own developments of dynamic charging.

Tongur sees a patchwork approach to charging in the medium term, with in-road charging playing a role within a larger infrastructure.

“I think having a shared charging platform that can charge any type of vehicle, not just trucks, … makes the greatest sense for us to meet the electrification demand that will come and that is coming,” Tongur said. “That’s why we need to start thinking and doing stuff right now.”

 

For more on our NEXT Series, click here, and tune in to Yahoo Finance every Monday at 10 a.m. ET.

Akiko Fujita is an anchor and reporter for Yahoo Finance. Follow her on Twitter @AkikoFujita.

Click here for the latest technology news that will impact the stock market.

Read the latest financial and business news from Yahoo Finance