Connect with us

Published

on

Sundar Pichai, CEO, Alphabet Inc., during the Google I/O developers conference in Mountain View, California, May 10, 2023.

David Paul Morris | Bloomberg | Getty Images

Google announced a new artificial intelligence model on Friday that can help it train robots to understand tasks like throwing out trash.

The Robotics Transformer 2 (RT-2) is a vision-language-action model trained on information and images on the internet that can be translated into actions for the robot, Google said in a blog post.

While a task like picking up the trash sounds simple to humans, it requires an understanding of a series of tasks for a robot to learn. For example, the robot must first be able to recognize what items constitute trash, then know to pick it up and throw it away. Rather than program a robot to do those specific tasks, RT-2 allows the robot to use knowledge from around the web to help it understand how to complete the task, even if it hasn’t been explicitly trained on the exact steps.

The new model nearly doubled the robots’ performance on previously-unseen scenarios, compared to the earlier version of the model, Google said. The new version can use rudimentary reasoning to respond to user commands, Google added.

The company doesn’t have imminent plans to widely release or sell robots with the new technology, The New York Times reported. But eventually, they could be used in warehouses or as home assistants, the Times added.

Subscribe to CNBC on YouTube.

WATCH: How autonomous inventory robots could save retailers billions

How autonomous inventory robots could save retailers billions

Continue Reading

Technology

Google criticized as AI Overview makes obvious errors, such as saying former President Obama is Muslim

Published

on

By

Google criticized as AI Overview makes obvious errors, such as saying former President Obama is Muslim

It’s been less than two weeks since Google debuted “AI Overview” in Google Search, and public criticism has mounted after queries have returned nonsensical or inaccurate results within the AI feature — without any way to opt out.

AI Overview shows a quick summary of answers to search questions at the very top of Google Search. For example, if a user searches for the best way to clean leather boots, the results page may display an “AI Overview” at the top with a multistep cleaning process, gleaned from information it synthesized from around the web.

But social media users have shared a wide range of screenshots showing the AI tool giving incorrect and controversial responses.

Google, Microsoft, OpenAI and other companies are at the helm of a generative AI arms race as companies in seemingly every industry rush to add AI-powered chatbots and agents to avoid being left behind by competitors. The market is predicted to top $1 trillion in revenue within a decade.

Here are some examples of errors produced by AI Overview, according to screenshots shared by users.

When asked how many Muslim presidents the U.S. has had, AI Overview responded, “The United States has had one Muslim president, Barack Hussein Obama.”

When a user searched for “cheese not sticking to pizza,” the feature suggested adding “about 1/8 cup of nontoxic glue to the sauce.” Social media users found an 11-year-old Reddit comment that seemed to be the source.

Attribution can also be a problem for AI Overview, especially in attributing inaccurate information to medical professionals or scientists.

For instance, when asked, “How long can I stare at the sun for best health,” the tool said, “According to WebMD, scientists say that staring at the sun for 5-15 minutes, or up to 30 minutes if you have darker skin, is generally safe and provides the most health benefits.”

When asked, “How many rocks should I eat each day,” the tool said, “According to UC Berkeley geologists, people should eat at least one small rock a day,” going on to list the vitamins and digestive benefits.

The tool also can respond inaccurately to simple queries, such as making up a list of fruits that end with “um,” or saying the year 1919 was 20 years ago.

When asked whether or not Google Search violates antitrust law, AI Overview said, “Yes, the U.S. Justice Department and 11 states are suing Google for antitrust violations.”

The day Google rolled out AI Overview at its annual Google I/O event, the company said it also plans to introduce assistant-like planning capabilities directly within search. It explained that users will be able to search for something like, “Create a 3-day meal plan for a group that’s easy to prepare,” and they’d get a starting point with a wide range of recipes from across the web.

“The vast majority of AI Overviews provide high quality information, with links to dig deeper on the web,” a Google spokesperson told CNBC in a statement. “Many of the examples we’ve seen have been uncommon queries, and we’ve also seen examples that were doctored or that we couldn’t reproduce.”

The spokesperson said AI Overview underwent extensive testing before launch and that the company is taking “swift action where appropriate under our content policies.”

The news follows Google’s high-profile rollout of Gemini’s image-generation tool in February, and a pause that same month after comparable issues.

The tool allowed users to enter prompts to create an image, but almost immediately, users discovered historical inaccuracies and questionable responses, which circulated widely on social media.

For instance, when one user asked Gemini to show a German soldier in 1943, the tool depicted a racially diverse set of soldiers wearing German military uniforms of the era, according to screenshots on social media platform X.

When asked for a “historically accurate depiction of a medieval British king,” the model generated another racially diverse set of images, including one of a woman ruler, screenshots showed. Users reported similar outcomes when they asked for images of the U.S. founding fathers, an 18th-century king of France, a German couple in the 1800s and more. The model showed an image of Asian men in response to a query about Google’s own founders, users reported.

Google said in a statement at the time that it was working to fix Gemini’s image-generation issues, acknowledging that the tool was “missing the mark.” Soon after, the company announced it would immediately “pause the image generation of people” and “re-release an improved version soon.”

In February, Google DeepMind CEO Demis Hassabis said Google planned to relaunch its image-generation AI tool in the next “few weeks,” but it has not yet rolled out again.

The problems with Gemini’s image-generation outputs reignited a debate within the AI industry, with some groups calling Gemini too “woke,” or left-leaning, and others saying that the company didn’t sufficiently invest in the right forms of AI ethics. Google came under fire in 2020 and 2021 for ousting the co-leads of its AI ethics group after they published a research paper critical of certain risks of such AI models and then later reorganizing the group’s structure.

In 2023, Sundar Pichai, CEO of Google’s parent company, Alphabet, was criticized by some employees for the company’s botched and “rushed” rollout of Bard, which followed the viral spread of ChatGPT.

Correction: This article has been updated to reflect the correct name of Google’s AI Overview. Also, an earlier version of this article included a link to a screenshot that Google later confirmed was doctored.

Continue Reading

Technology

Nvidia CEO Jensen Huang’s net worth swells from $3 billion to $90 billion in five years

Published

on

By

Nvidia CEO Jensen Huang's net worth swells from  billion to  billion in five years

Jensen Huang, co-founder and CEO of Nvidia, during the Nvidia GPU Technology Conference in San Jose, California, on March 19, 2024.

David Paul Morris | Bloomberg | Getty Images

Five years ago, Nvidia CEO Jensen Huang owned a stake in his chipmaker worth roughly $3 billion. After Thursday’s rally, which pushed the stock to a record, his holdings now stand at more than $90 billion.

Nvidia late Wednesday reported first-quarter earnings that topped estimates, with sales jumping more than 200% for a third straight quarter, driven by demand for artificial intelligence processors.

Huang also delivered a better-than-expected forecast and indicated to investors that the company sees insatiable demand for its AI graphics processing units, or GPUs. The company signaled its customers, especially the big cloud companies, could get a strong return on their investment in the pricey chips.

“We are fundamentally changing how computing works and what computers can do,” Huang said.

Huang owns about 86.76 million shares of Nvidia, or more than 3.5% of the company’s outstanding shares. With the stock rising over 9% to close at a price of nearly $1,038 per share on Thursday, the value of his stake rose by about $7.7 billion.

Nvidia shares have more than doubled this year after tripling in 2023. They are up about 28-fold in the past five years. Huang added shares to his stake in 2022, when the stock hit relative lows before the AI boom.

Huang, 61, founded the Silicon Valley company in 1993 to build GPUs for 3D gaming. While gaming was the company’s biggest business for decades, Nvidia has dipped into other markets, including cloud gaming subscriptions, the metaverse and cryptocurrency mining chips.

But Nvidia’s fortunes shifted dramatically in late 2022, when OpenAI released ChatGPT, opening up the concept of generative AI to the broader public. The technology showcased a future in which computers won’t just retrieve new information from databases, but can also generate new content and answers to questions from large caches of unsorted data.

OpenAI does most of its AI development on Nvidia GPUs. As other companies such as Microsoft, Google and Meta bolstered their investments in AI research and development, they needed billions of dollars worth of the latest AI chips to build out their models.

Huang has been the face of Nvidia and its principal salesperson, constantly extolling the potential and power of using the company’s GPUs for building AI.

Nvidia, which has been developing AI software and tools for more than a decade, ended up in prime position to become the top supplier to the biggest technology companies. The company now has about 80% of the market for AI chips, and Huang is among the 20 richest people in the world.

Don’t miss these exclusives from CNBC PRO

Executive Edge: Nvidia CEO pay rises to $34.2 million

Continue Reading

Technology

Here’s what it’s like inside the operating room when someone gets a brain implant

Published

on

By

Here's what it's like inside the operating room when someone gets a brain implant

Dr. Joshua Bederson places Precision Neuroscience’s electrodes onto a brain.

Ashley Capoot

As the lights dimmed in an operating room at The Mount Sinai Hospital in New York City, Dr. Joshua Bederson prepared to make history.

Bederson, system chair for the Department of Neurosurgery at Mount Sinai Health System, is no stranger to long hours in an operating room. The former competitive gymnast has completed more than 6,500 procedures in his career, and he said he visualizes the steps for each one as if he’s rehearsing for a routine.   

On this particular morning in April, Bederson was readying for a meningioma resection case, which meant he would be removing a benign brain tumor. Bederson said his primary focus is always on caring for the patient, but in some cases, he also gets to help advance science. 

This procedure was one such case. 

A small crowd gathered as Bederson took his seat in the operating room, his silhouette aglow from the bright white light shining on the patient in front of him. Health-care workers, scientists and CNBC craned forward – some peering through windows – to watch as Bederson placed four electrode arrays from Precision Neuroscience onto the surface of the patient’s brain for the first time. 

An electrode is a small sensor that can detect and carry an electrical signal, and an array is a grid of electrodes. Neurosurgeons use electrodes during some procedures to help monitor and avoid important parts of the brain, like areas that control speech and movement.

Precision is a three-year-old startup building a brain-computer interface, or a BCI. A BCI is a system that decodes neural signals and translates them into commands for external technologies. Perhaps the best-known company in the field is Neuralink, which is owned by Tesla and SpaceX CEO Elon Musk.

Other companies like Synchron and Paradromics have also developed BCI systems, though their goals and designs all vary. The first application of Precision’s system will be to help patients with severe paralysis restore functions like speech and movement, according to its website. 

Stephanie Rider of Precision Neuroscience inspects the company’s microelectrode array

Source: Precision Neuroscience

Precision’s flagship BCI is called the Layer 7 Cortical Interface. It’s a microelectrode array that’s thinner than a human hair, and it resembles a piece of yellow scotch tape. Each array is made up of 1,024 electrodes, and Precision says it can conform to the brain’s surface without damaging any tissue.

When Bederson used four of the company’s arrays during the surgery in April, he set a record for the highest number of electrodes to be placed on the brain in real-time, according to Precision. But perhaps more importantly, the arrays were able to detect signals from the patient’s individual fingers, which is a far greater amount of detail than standard electrodes are able to capture.

Using Precision’s electrode array is like turning a pixilated, low-resolution image into a 4K image, said Ignacio Saez, an associate professor of neuroscience, neurosurgery and neurology at the Icahn School of Medicine at Mount Sinai. Saez and his team oversee Precision’s work with Mount Sinai.

“Instead of having 10 electrodes, you’re giving me 1,000 electrodes,” Saez told CNBC in an interview. “The depth and the resolution and the detail that you’re going to get are completely different, even though they somehow reflect the same underlying neurological activity.”

Bederson said accessing this level of detail could help doctors be more delicate with their surgeries and other interventions in the future. For Precision, the ability to record and decode signals from individual fingers will be crucial as the company works to eventually help patients restore fine motor control. 

The data marks a milestone for Precision, but there’s a long road ahead before it achieves some of its loftier goals. The company is still working toward approval from the U.S. Food and Drug Administration, and it has yet to implant a patient with a more permanent version of its technology. 

“I think these are little baby steps towards the ultimate goal of brain-computer interface,” Bederson told CNBC in an interview.

Inside the operating room

Dr. Joshua Bederson prepares for surgery at The Mount Sinai Hospital.

Ashley Capoot

Bederson’s surgery in April was not Precision’s first rodeo. In fact, it marked the 14th time that the company has placed its array on a human patient’s brain. 

Precision has been partnering with academic medical centers and health systems to perform a series of first-in-human clinical studies. The goal of each study varies, and the company announced its collaboration with Mount Sinai in March. 

At Mount Sinai, Precision is exploring different applications for its array in clinical settings, like how it can be used to help monitor the brain during surgery. In these procedures, surgeons like Bederson temporarily place Precision’s array onto patients who are already undergoing brain surgery for a medical reason. 

Patients give their consent to participate beforehand. 

It’s routine for neurosurgeons to map brain signals with electrodes during these types of procedures. Bederson said the current accepted practice is to use anywhere between four to almost 100 electrodes – a far cry from the 4,096 electrodes he was preparing to test. 

Electrode arrays from Precision Neuroscience displayed on a table.

Ashley Capoot

Precision’s arrays are in use for a short portion of these surgeries, so CNBC joined the operating room in April once the procedure was already underway. 

The patient, who asked to remain anonymous, was asleep. Bederson’s team had already removed part of their skull, which left an opening about the size of a credit card. Four of Precision’s arrays were carefully laid out on a table nearby.

Once the patient was stabilized, Precision’s employees trickled into the operating room. They helped affix the arrays in an arc around the opening on the patient’s head, and connected bundles of long blue wires at the other end to a cart full of equipment and monitors.

Dr. Benjamin Rapoport, Precision’s co-founder and chief scientific officer, quietly looked on. Every major procedure presents some risks, but the soft-spoken neurosurgeon’s calm demeanor never wavered. He told CNBC that each new case is just as exciting as the last, especially since the company is still learning. 

Experts help set up the wiring for Precision Neuroscience’s technology.

Ashley Capoot

Bederson entered the operating room as Precision’s preparations neared their end. He helped make some final tweaks to the set up, and the overhead lights in the operating room were turned off. 

Ongoing chatter quieted to hushed whispers. Bederson was ready to get started. 

He began by carefully pulling back a fibrous membrane called the dura to reveal the surface of the brain. He laid a standard strip of electrodes onto the tissue for a few minutes, and then it was time to test Precision’s technology. 

Using a pair of yellow tweezers called long bayonet forceps, Bederson began placing all four of Precision’s electrode arrays onto the patient’s brain. He positioned the first two arrays with ease, but the last two proved slightly more challenging. 

Bederson was working with a small section of brain tissue, which meant the arrays needed to be angled just right to lay flat. For reference, imagine arranging the ends of four separate tape measures within a surface area roughly the size of a rubber band. It took a little reconfiguring, but after a couple of minutes, Bederson made it happen.

Real-time renderings of the patient’s brain activity swept across Precision’s monitors in the operating room. All four arrays were working.  

In an interview after the surgery, Bederson said it was “complicated” and “a little bit awkward” to place all four arrays at once. From a design perspective, he said two arrays with twice as many points of contact, or longer arrays with greater spacing would have been helpful.  

Bederson compared the arrays to spaghetti, and the description was apt. From where CNBC was watching, it was hard to tell where one stopped and the next began.  

Once all the arrays were placed and actively detecting signals, Precision’s Rapoport stood with his team by the monitors to help oversee data collection. He said the research is the product of a true team effort from the company, the health system and the patient, who often doesn’t get to see the benefits of the technology at this stage. 

“It takes a village to make this sort of thing move forward,” Rapoport said. 

CNBC left the operating room as Bederson began removing the tumor, but he said the case went well. The patient woke up afterward with some weakness in their foot since the surgery was within that part of the brain, but Bederson said he expected the foot would recover in around three to four weeks. 

Employees from Precision Neuroscience collecting data.

Ashley Capoot

Rapoport was present at this particular surgery because of his role with Precision, but he’s well acquainted with the operating rooms at Mount Sinai. 

Rapoport is a practicing surgeon and serves as an assistant professor of neurosurgery at the Icahn School of Medicine at Mount Sinai. Rapoport reports to Bederson, and Bederson said the pair have known one another since Rapoport was in residency at Weill Cornell Medicine.

Dr. Thomas Oxley, the CEO of the competing BCI company Synchron, is also a faculty member under Bederson. Synchron has built a stent-like BCI that can be inserted through a patient’s blood vessels. As of early February, the company had implanted its system into 10 human patients. It is also working toward FDA approval. 

Bederson has an equity stake in Synchron, but he told CNBC he didn’t realize how much it would prevent him from participating in research with the Synchron team. He has no monetary investment in Precision. 

“I really did not want to have any financial interest in Precision because I think it has an equally promising future and wanted to advance the science as fast as I could,” Bederson said. 

Rapoport also helped co-found Musk’s Neuralink in 2017, though he departed the company the following year. Neuralink is building a BCI designed to be inserted directly into the brain tissue, and the company recently received approval to implant its second human patient, according to a report from The Wall Street Journal on Monday. 

As the BCI industry heats up, Bederson said the amount that scientists understand about the brain is poised to “explode” over the next several years. Companies like Precision are just getting started. 

Dr. Joshua Bederson helps set up Precision Neuroscience’s electrode arrays.

Ashley Capoot

“I really feel like the future is where the excitement is,” Bederson said.

Rapoport said Precision is hoping to receive FDA approval for the wired version of its system “within a few months.” This version, which is what CNBC saw in the operating room, would be for use in a hospital setting or monitored care unit for up to 30 days at a time, he said. 

Precision’s permanent implant, which will transmit signals wirelessly, will go through a separate approval process with the FDA. 

Rapoport said Precision hopes to implant “a few dozen” patients with the wired version of its technology by the end of the year. That data collection would give the company a “very high level of confidence” in its ability to decode movement and speech signals in real-time, he said. 

“Within a few years, we’ll have a much more advanced version of the technology out,” Rapoport said.

Continue Reading

Trending