New photosynthesis study moves research in the ‘light’ direction

I’ll riddle you this: What do plants eat when they're kind of hungry but not that hungry?

A 'light' snack.  (Don’t worry, I’m not giving up the day job any time soon.)

Every year, millions of schoolchildren are taught about photosynthesis. For those who aren’t up on their fourth-grade science, it’s the process through which plants turn sunlight into energy. However, as fundamental as the concept may seem, there are still many unanswered questions regarding how it truly works. These include the photophysics of the process and the molecular changes that take place when a plant absorbs sunlight.

We may be one step closer to understanding these processes thanks to new research conducted by the Cavendish Laboratory at Cambridge University.

Why should we care? For starters, some of the information discovered could make a big difference in solar-energy production. If we can figure out exactly how photosynthesis works, we just might be able to mimic how the plant ecosystem harvests sunlight to create more efficient and possibly cheaper solar panels.

But the reason why understanding photosynthesis is so difficult is because it occurs at speeds that are too fast for traditional monitoring systems. How fast? Try a millionth of a millionth of a second. In the Cambridge study, published in Nature this March, researchers employed ultra-rapid spectroscopic techniques — laser pulses directed at live cell samples — to try to observe these biological changes.

This method allowed researchers to "follow the flow of energy in the living photosynthetic cells on a femtosecond scale — a thousandth of a trillionth of a second," and carefully see what the electrons are doing.

“We didn’t know as much about photosynthesis as we thought we did, and the new electron transfer pathway we found here is completely surprising,” Cambridge’s Jenny Zhang, who coordinated the research, said in a statement.

 “Achieving this would open up many exciting possibilities where photosynthetic cells and their components can serve as self-generating, self-repairing catalysts,” she told Gizmodo.

OK … Let’s see some ‘I.D.’

Watch out, Tesla because Volkswagen wants its new EV sedan to be top of mind when it comes to car buyers who are finally taking the electrified plunge. VW hopes to entice customers with the 2025 ID.7, which it showed off at the Shanghai Auto Show this week.

The ID.7’s size is planted firmly between Tesla's Model S and Model 3 sedans. If equipped with a 86kWh battery, it can drive 435 miles on a full charge, or 382 miles with the standard 77kWh cell. It will reach 60 mph in the low 5-second range.

“Our focus continues to be SUVs, but obviously the sedan market represents 16% of the total market in the U.S.,” Pablo Di Si, president and CEO of Volkswagen of America, told Yahoo Finance. “And bringing the ID.7 … in 2024 it is the right approach to take for VW.”

The ID.7 is part of a massive rush to bring new EVs to the market, as Ford, GM, Hyundai and many other automakers try catching up to Tesla and meet growing demand for these gasoline teetotalers.

In a design first, VW claims that the ID.7 will come equipped with augmented-reality technology that would project information in 3D on the windshield, giving the vehicle a standout feature that not even Tesla can match.

Other significant interior features of the new ID.7 include an electrochromic glass roof that can be instantly changed from clear to tinted.

“By 2030, we're going to launch 25 electric vehicles for the Volkswagen Group family, and we aim to get 10% market share in the US by 2030,” Di Si told Yahoo. “So, I think we are at the right time with the right products in the U.S., and we need to have a really huge product offensive.”

Sales of the ID.7 will start in China later this year, followed by the United States in 2024.

Cutting our carbon footprint from Mount Olympus

Looks like a Greek god will be entering the world of deep learning. Many of our favorite web services, from Amazon to ChatGPT to Netflix, run using artificial intelligence. But training that AI to work consumes a lot of energy.

Enter Zeus, a new framework created by the University of Michigan to optimize energy use in AI training. Using Zeus, companies could cut their AI training energy consumption by 75%, and without any new hardware, reports the American Association for the Advancement of Science.

According to Mosharaf Chowdhury, whose team created Zeus, training AI chatbots are a drain on the grid. "At extreme scales, training the GPT-3 model just once consumes 1,287 MWh, which is enough to supply an average U.S. household for 120 years."

Jae-Won Chung, who was part of Chowdhury’s team, says that existing AI training work focuses mostly on optimizing for speed, often without thinking about how that affects energy efficiency. "We found that the energy we put into GPUs doesn't pay off as well as it used to, which means we can use a lot less energy without slowing down too much," he said.