Elon Musk has announced that Tesla will likely begin retrofitting its new, more powerful processing chip into older vehicles before the end of the year. Tesla claims that the new FSD chip, which was developed in-house, performs 21 times better than the previous Nvidia chips. The new chips cost 20 percent less than the Nvidia chips while only consuming a little more power. Newer Model S, X, and 3 cars have been shipping with the new chip since before its announcement. Around 500,000 existing cars are currently compatible with the new chip, and Tesla owners will receive the upgrade for free if they have purchased the Full Self-Driving add-on package.
A new anti-bullying measure by Instagram uses artificial intelligence to flag comments. Account owners can also now shadowban users who post on their comments, which means that only the commenter and account owner will be able to see the messages. Once the AI detects that a user is posting a comment that may be deemed offensive or bullying, the user is prompted with a message to confirm whether they really want to post the comment. In tests, this system was shown to lower the number of offensive comments posted. Instagram has been implementing much needed anti-bullying measures for a while now. However, studies have shown that many teens create new accounts entirely for bullying which makes it more difficult to stop.
Scientists at Durham University have demonstrated that there may be another valid explanation for gravity that would still be compatible with the way our universe works. Albert Einstein’s theory of General Relativity is the foundation of modern cosmology and plays a major role in our everyday life. The Chameleon Theory is an alternative to General Relativity which says that gravity changes according to the environment. Simulations run by the team at Durham demonstrate that the same galaxies would form in the universe regardless of which theoretical model was used. These simulations will be further tested by observations using the Square Kilometer Array telescope that will begin operating in 2020.
A ‘Vegebot’ developed by a team at the University of Cambridge is able to harvest iceberg lettuce using machine learning. Crops such as potatoes and wheat have been harvested using machinery for decades. However, iceberg lettuce presents a more difficult problem as it is harder to determine when they are ripe, and the crops are easily damaged during harvesting. The Vegebot is trained on images of ripe crops and is able to tell which crops are ready for harvest with a visual scan. A second component focuses on harvesting the crop without damage so that it is supermarket ready. The process usually takes a few seconds for a human worker, but the farming industry is currently facing a worker shortage. These bots can also work around the clock, going through fields multiple times to harvest exactly at the right time.
Python autocomplete is a project that experiments with using an LSTM model to autocomplete python code. Test results show that it can save users between 30 to 50 percent of keystrokes. It is currently set to predict up to 10 characters ahead. A saved model that is trained on TensorFlow/models is available on the repository.
Tech workers have been participating in a lot of activism over the last few years, with some results. Google employees have affected how artificial intelligence is to be used, and employees from other companies have protested successfully to make changes to their workplace. However, employees have failed to organize unions. There are many factors preventing unionization from happening. Some employees may feel too close to their employers due to the small size of a company, or highly paid employees may see themselves as independent contractors with leverage who don’t need union support. Some tech companies have dismissed employees who tried to organize unions in the past. The culture in Silicon Valley is changing, and unions may be the best way to stop employees from being unfairly pressured by their companies.
Get the most important tech, science, & coding news in a free daily email. Read by +1,250,000 software engineers and tech workers.