Artificial Intelligence - Page 69

My name is Ganesh [Venkataramanan, Tesla director and Dojo boss] and I lead Dojo. It's an honor to present this project on behalf of the multi-disciplinary Tesla team that is working on this project.

As you saw from Milan, there's an insatiable demand for speed, actualized capacity for neural network training -- and Elon prefetched this, and a few years back he asked us to design a super-fast training computer, and that's how we started Project Dojo.

Our goal is to achieve best AI training performance and support all these larger, more complex models that Andre's team are dreaming of and be power-efficient, and cost-effective at the same time. So we thought about how to build this, and we came up with a Distributed Compute Architecture.

The Pentagon is using an experimental artificial intelligence program that allows it to see "days in advance" and look into the future.

The new experimental program is called the Global Information Dominance Experiments (GIDE), and it combines artificial intelligence with cloud computing and large data pools. The newest version of GIDE is GIDE 3, and according to General Glen VanHerck, the commander of the US Northern Command, the idea behind the new program is to "achieve information dominance" and "decision making superiority".

VanHerck stated to the press conference that the Pentagon is currently living in a "reactive environment" where it will respond to rival nations' actions. Now, VanHerck says, "What we've seen is the ability to get way further what I call left of being reactive to actually being proactive. And I'm talking not minutes and hours, I'm talking days."

A bipedal robot named Cassie has achieved what no other robot has achieved -- successfully completing a 5K run on a single charge.

Oregon State University engineers have managed to teach Cassie to run using a deep reinforcement learning algorithm. The project is led by Agility Robotics who is attempting to combine robot controls with machine learning tools, and as a result of the engineers' efforts, we are presented with Cassie. The team of engineers said that Cassie taught itself to upright, which meant that it didn't require any tethering device to stabilize it as it ran.

Cassie ran around the Oregon State University campus for five kilometers or about 3.11 miles, and during the test, the running robot fell down twice, once due to a computer overheating and a second time when it when around a corner too fast. Cassie completed the five-kilometer run in 53 minutes, 3 seconds, and according to Jeremy Dao, a project team member, they were able to "reach the limits of the hardware and show what it can do."

Why use humans to build next-gen chips when AI is so much freaking faster?

That's what Google is doing, with the everything giant saying that it has created artificial intelligence (AI) software that is so powerful it can design next-gen computer chips much faster than any humans can.

In a new paper published in the journal Nature, the authors of the paper explain that it takes humans months and months to design a new chip -- meanwhile, AI takes less than 6 hours to do the same thing. Azalia Mirhoseini and Anna Goldie are two of the authors behind the new paper, who are co-heads of machine learning for systems at Google. They explained: "Our method has been used in production to design the next generation of Google TPU".

Artificial intelligence (AI) now has the capabilities to predict space weather that is caused by our Sun accurately.

Researchers from the University of Graz have created a new neural network that allows for artificial intelligence to reliably predict changes in the Sun's coronal holes from space-based observations. As you already know, the light emitted from the Sun plays a vital role in our existence here on Earth. Additionally, the light from the Sun interacting with Earth's magnetic field can influence our electronics, and in extreme cases, when the Sun blasts Earth with too many charged particles, our electricity grids can be temporarily knocked offline by geomagnetic storms.

Now, the researchers have developed a new neural network that examines some of the dark regions on the Sun called coronal holes. The location and size of coronal holes depend on how much solar activity is occurring on the Sun. Researchers are tasked with observing these coronal holes and their movements to predict space weather, which can then affect us here on Earth.

I didn't think I'd see this today, yet here we are -- Pornhub has announced its Remastured Film Library, with the porn giant using AI and deep learning to restore some of the oldest erotic films ever made.

Pornhub used the power of AI and machine learning letting it watch over 100,000 adult videos and images from Pornhub's swelling library, giving their AI a "sexual education" and "teaching it a new dirty trick". That new dirty trick is the ability to colorize vintage porn for a modern audience.

The site has even colorized the first kiss ever captured on film, with The Kiss dating back to 1886 -- 135 years ago now. The 18-second movie was "denounced as shocking and obscene to viewers" at the time, so just imagine what those same people would say at some of the videos you can find online in 2021. But now, The Kiss is in 4K 60FPS thanks to AI and deep learning.

A company has announced that it will be using deepfake technology to create a synthetic voice that celebrities and other prominent figures can license out.

Veritone has said it plans to use its Marvel.au platform to monetize deepfakes of celebrities, politicians, and any other public figure. The AI-enabled synthetic voice can be licensed out by the celebrity to companies, brands, and influencers. By having a deepfake of a celebrity's voice media companies will be able to "accelerate and amplify content creation", according to Veritone.

Essentially, Veritone would create a deepfake, or a clone of a celebrity's voice that can then be bought by companies to do brand duties that the celebrity would get paid for. Veritone mentions that creating a voice clone for a person to perform brand deals would remove the physical need for that person to on set to shoot it, which could raise the ceiling for that celebrity's total monetized content output. For more information on this story, check out this link here.

Hyundai Motor Group announced its intent to purchase 80% of Boston Dynamics from SoftBank. The deal values the company at $1.1 billion. Hyundai intends to use the technology"to realize progress for humanity."

Hyundai Motor Group is doubling down on its robotics manufacturing portfolio. The company already designs and manufactures robots for automotive assembly lines. With the acquisition of Boston Dynamics, the company plans to develop robots for non-industrial use cases.

Boston Dynamics, famous for its videos of its robotic dog named Spot, builds some of the world's most advanced robotics technology. The company, which derived its technology from DARPA research, builds quadrupedal and bipedal robots that can walk like the creatures they mimic. The company also makes AI algorithms to enable these robots to adapt to their environment.

You might not have heard of Parler, but it's an alterate social media platform that is similar to Facebook and Twitter but calls itself "viewpoint neutral".

Well, Parler use has exploded post-US elections on November 3, 2020 -- with Parler being the #1 app on both the Apple iOS and Google Play app stores with over 1 million downloads in the days after the election. People wanted a new home where they weren't being censored, with many popular conservatives moving from Facebook and Twitter onto Parler, driving up usage.

Parler opened in 2018 by John Matze and Jared Thomson who were sick of the "ideological suppression" by the technology companies in the US. But the number Parler are pulling in are significant, with Matze saying Parler added 4.5 million in a span of just 5 days.

A new manga has just dropped, with it being the first that was plotted and designed by artificial intelligence (AI). The AI itself looked at, and learned the artistic style of Astro Boy creator Osamu Tezuka.

The AI-made manga is called "PHAEDO" and was released in the weekly comic magazine "Morning" reports the Japan Times. The AI generated character designs and basic story lines from the inspiration of the Astro Boy creator, and then it was handed to professional creators to add in things like clothing and dialogue.

Kioxia Corporation has announced that it is releasing the English version of the manga, as part of TEZUKA 2020 and the first phase of the company's new #FutureMemories brand campaign. The English version of PHAEDO will be released on Kioxia's own TEZUKA 2020 website.

The coronavirus is undoubtedly a terrible thing to have happened to everyone, but out of the world's misery shines fantastic examples of humans coming together for one common cause.

One of those common causes is Folding@Home, which is a computing project run by a team at Stanford University. The whole premise of Folding@Home is that the public can donate spare computation power from either their GPU or CPU to the project, and then the project aims that computational power at calculations that give scientists more understanding about a topic, or subject. At the moment, Folding@Home is aiming its power at understanding the coronavirus, and it has recently hit a new power level - 2.4 exaFLOPS.

In the above Twitter post from the official Folding@Home Twitter account, it's stated that the collective power of the Folding@Home project is almost at 2.5exaFLOPS, which is "faster than the top 500 supercomputers combined!". The Twitter post also contains a graph, and as you can see, Folding@Home is clearly exceeding all competing supercomputers, even to the point that it's "15x faster than any current supercomputer". It's absolutely fantastic to see the technology community rally together under one great cause. If you want to check out more about Folding@Home, or would like to donate some of your own computation power, check out this link here.

Artificial Intelligence (AI) has some concerns around it, and one of the most prominent names voicing those concerns if SpaceX and Tesla CEO Elon Musk.

If you didn't know, Musk founded a non-profit organization called OpenAI with Sam Altman, Ilya Sutskever, Greg Brockman, Wojciech Zaremba, and John Schulman. The goal of the organization was to develop new and safe was to monitor and regulate AI development. Musk has since left the organization but is still voicing his concerns over the oversight on AI development.

Musk took to Twitter in response to a new MIT Technology Review profile of OpenAI. Here's what he said, "OpenAI should be more open imo (in my opinion). All orgs developing advanced AI should be regulated, including Tesla". Judging on Musk's responses, it can be assumed that Musk is now somewhat distant from the organization's operations. Back in 2019, OpenAI started a for-profit arm owned by the non-profit parent company, and also accepted a $1 billion investment from Microsoft.

Deepfake technology is getting uncomfortably good, and while most of what you have probably seen from deepfakes is movie-related, check out this video game one.

Above we have a video from BabyZone who used Deepfake technology to include various celebrities in Mortal Kombat 11. Using the deepfake tech, BabyZone was able to successfully replace Sub-Zero with Keanu Reeves from John Wick, Terminator with Bruce Campbell, Kabal as Dwayne 'The Rock' Johnson, Johnny Cage as Jean-Claude Van Damme, Kung Lao with Jackie Chan, Liu Kang with Bruce Lee and finally The Joker with Joaquin Phoenix.

I think Keanu Reeves as Sub-Zero and Jackie Chan as Kung Lao are the two that are particularly good out of the set. Unfortunately, you cannot download these to play as BabyZone used his own facesets, and the footage that you are seeing was edited in post. Regardless of not being able to download them, these deepfakes are extremely impressive.

It was only last week I reported on some law enforcement adopting a new artificial intelligence to assist them in investigations. Now, the company behind that artificial intelligence has been asked to cease and desist by multiple social media platforms.

The artificial intelligence we are talking about here is from Clearview and is a facial recognition software that uses public personal images scraped from multiple social media platforms to identify people. This AI has been fed over 3 billion publically available images and can now identify almost anyone at the drop of a hat. Clearview, the company behind the AI, has been pushing into government and law enforcement as they believe what they have created can assist in investigations.

According to a Facebook spokesperson who spoke to Buzzfeed News, Facebook has sent "multiple letters" asking Clearview to cease and desist scaping "data, images and media" from Facebook and Instagram. Facebook isn't alone in asking Clearview to 'please stop', as back in late January, Twitter also sent letters to Clearview. YouTube and Venmo are also in the same letter sending boat as the other social media platforms.

Chicago Police are using an artificial intelligence that could be in clear breach of people privacy online, as it scans everyone's social media.

The artificial intelligence is called Clearview AI and it is a massive database that includes 3 billion photos taken from social media and other platforms such as Facebook, YouTube and Twitter. Users of the AI feed an image of the person they want to look for into its system then the AI will cycle through its database and present the users with a bunch of different images from different platforms that 'match' the fed image. At the moment the AI is being used by the FBI, Homeland Security and the Chicago Police Department (CPD).

While this AI system would definitely save law enforcement some time in tracking criminals, many privacy advocates have said that it is in complete violation of peoples privacy rights. New Jersey's attorney general Gurbir Grewal, said "Until this week, I had not heard of Clearview AI. I was troubled." According to ACLU of New Jersey, Grewal "put a moratorium on Clearview AI's chilling, unregulated facial recognition software."

CES 2020 - American Airlines will be using Google Assistant's interpreter mode as a means of making travelers more comfortable in lounges.

If you weren't aware, Google Assistant has a really cool Interpreter Mode that allows for people to communicate over the language barrier. Engadget managed to spot Google Assistant's Interpreter Mode at the Los Angeles International Airport's Admirals Club lounge, where American Airlines was testing it out on Google Nest Hubs.

Interpreter Mode can currently translate 29 different languages in real-time; those languages are the following: Arabic, French, German, Japanese, Russian, Spanish, and Vietnamese. Engadget has also said that according to American Airlines Interpreter Mode will only be used if a multilingual team member isn't present to assist travelers.

CES 2020 - If you are a parent, a baby monitor is an essential thing to have around you when you are trying to relax. So why not get a baby monitor that can do it all?

KamiBaby seems to be the cream of the crop when it comes to baby monitors, as this small cute device can do it all when it comes to giving you everything you need to know about your baby while he/she is sleeping. So how is this baby monitor different from the one you already have?

KamiBaby uses a neural network AI to monitor everything your baby is doing; it can read breathing, magnify motion when it happens, has a high-resolution camera, gives sleep summary's, temperature readings, has crying detection, and more. The invention of KamiBaby originated from an Indiegogo fundraiser that accumulated over $62,000. YI Technology, the company behind the invention, is giving people a 50% discount on pre-orders priced at $99. When KamiBaby hits stores, it will be sold off the shelf for $199.

CES 2020 - Experience and AI-driven orgasm with the Lioness Generation 2, the latest vibrator that uses massive pools of data to give you the best experience possible.

The Lioness Generation 2 is a new vibrator that uses an astonishing 30,000 individual orgasms to deliver the user the best experience yet. It was shown off at this year's CES, and according to the slide above, "knowledge is power. And pleasure. Have better orgasms more often with AI-assisted guidance, backed by years of research and analysis of over 30,000 orgasms."

Not only does Generation 2 have a massive data set to work off of, but it also tracks a user's individual session, saves the data, and then relays tips back to the user. Of course, this vibrator wouldn't give the full experience unless it had an app to go along with it -- which is does. The app shows off all the previously mentioned data in graphs and gives further details on how to improve pleasure.

CES 2020 - Samsung has been teasing something about 'Project Neon' for a while now, and while some thought it would be some fancy full artificial human being -- in reality, it seems it's just an ultra-realistic human avatar.

Neon would be used for entertainment and business use, so think of a virtual person greeting you at a hotel, a receptionist, or even as an actor in guiding someone through a building or tutorial. We don't know what to expect exactly, but the head of Neon and computer-human interaction researcher Paranav Mistry tweeted a tease above.

He said that the company's new "Core R3" technology is capable of "autonomously create new expressions, new movements, new dialog (even in Hindi), completely different from the original captured data".

AI is being used for one of the best causes on this planet -- cancer. DeepMind, the Google-owned and UK-based artificial intelligence company is now using AI towards breast cancer detection.

Breast cancer detection isn't a perfect science, as there are a high number of false positives (this is when a doctor or expert sees a mammogram and says there's an issue, like breast cancer is spotted, but there is, in fact, no cancer). AI on the other hand, isn't having as many issues with false positives.

DeepMind trained its AI using de-identified data from patients in the US and UK, where it saw false positives reduced by 5.7% and false negatives by 9.% in the US. But in the UK the numbers were much different, with 1.2% and 2.7% respectively. The AI in question only had access to the most recent mammogram of each patient, but it was able to make screening decisions with higher accuracy than their human counterparts.