Yann LeCun, Meta’s chief AI scientist, said, “In terms of underlying techniques, ChatGPT is not particularly innovative.” According to sources, he was speaking at a small gathering of press and executives on Zoom last week.
“It’s nothing revolutionary, although that’s the way it’s perceived in public,” said LeCun. “It’s just that, you know, it’s well put together, it’s nicely done.”
LeCun said that such data-driven AI systems had been built by many research labs and companies in the past. He said the idea of OpenAI being the only one in its type of work is not correct.
LeCun said, “OpenAI is not particularly an advance as compared to the other labs, at all.” LeCun noted the several ways in which ChatGPT, and the program upon which it has been built, OpenAI’s GPT-3, is composed of many pieces of technology that have been developed over many years by different parties.
LeCun was invited to speak for an hour-and-a-half talk that was hosted by the Collective Forecast. This online, interactive discussion series is organized by Collective, which calls itself an AI platform designed to optimize the B2B sales.
In 2017, Microsoft had stepped in to acquire the platform after AltspaceVR decided to shut itself down. Now, the company says it will shut down the same service on March 10, 2023.
The tech giant will shift its attention to support immersive experiences that are powered by Microsoft Mesh, which is a platform that enables presence and shared experiences from anywhere, on any device, through mixed reality applications.
“With Mesh, we aspire to create a platform that offers the widest opportunity to all those involved, including partners, creators, and customers,” Microsoft said.
Moreover, in the coming term, the company will focus its VR efforts on workplace experiences, learning alongside and from its early partners and customers, and ensuring that it delivers a foundation that enables security, trust, and compliance.
Boston Dynamics’ Atlas robot is now capable of running, jumping, picking up and throwing heavy objects, and performing complex moves that may one day help humanity in factories or construction sites.
In a newly released video called “Atlas Gets a Grip,” the humanoid robot grabs and throws a bag of tools to a worker waiting at the top of a multi-story scaffold.
To some, the new skills may not seem as jaw-dropping since, in the previous videos, the robot dances to the song ‘Do you love me?’ and performs a backflip. However, Boston Dynamics says the technology used in the new control capabilities is more difficult to master than extreme acrobatics.
In the new video, Atlas can be seen picking up not only the tool bag but also a large wooden plank, running up stairs, jumping between levels, and pushing a big wooden box off the platform without losing balance, which according to the company is a “deceptively complex task.”
Boston Dynamics mentioned that for a robot, performing these manipulation tasks needs a more nuanced understanding of its environment, as compared to dancing, where the robot is blind to its surroundings.
Atlas, which is about 1.52 m tall and weighs 86 kg, has a total of 28 hydraulic joints, three onboard computers, and battery-powered electric motors for movement.
Mercedes-Benz has received permission to launch its autonomous driving system DRIVE PILOT in Nevada, reported Reuters. This makes the company the first automaker with a Level 3 system approved for regular use anywhere in the US.
“Mercedes has certified that their technology meets the Nevada ‘minimal risk condition’ requirement that needs a Level 3 or higher ‘fully autonomous’ vehicles that can stop if there is a certain malfunction in the system,” the Nevada DMV said.
A driver is still needed behind the wheel when DRIVE PILOT is engaged. Although under specific conditions, that driver doesn’t have to keep their eyes on the road. For instance, during highway driving, where traffic density is more, and the vehicle is traveling at speeds below 37 mph.
“This gives customers back time to focus on certain secondary tasks, such as communicating with colleagues via In-Car Office, browsing the internet, or relaxing while watching a movie,” writes Mercedes-Benz.
Drive Pilot needs the installation of certain hardware and cannot be retrofitted to the existing cars. Therefore, anyone interested in using the service in Nevada must be prepared to buy a new vehicle and hop up for the option.
An internal slideshow from Google suggests the company could announce as many as 20 new products this year alongside the new AI variant of Google Search, according to a report in New York Times.
The company is also looking to employ a fast-track review process, the ‘Green Lane’ initiative, to make sure the AI features are ready according to schedule. Google may even allow the AI teams to review their own work to ‘recalibrate’ the risk involved.
For the anticipated AI demo, Google is apparently looking to minimize misinformation, improve accuracy, and ensure safety by blocking specific words. As per the report, Google CEO Sundar Pichai invited the founders, Larry Page and Sergey Brin, for a special meeting where the trio reportedly discussed the AI chatbot’s rise.
There have been arguments about the threat that OpenAI’s ChatGPT poses to Google’s search engine, if not now, then in the future. The report by The New York Times suggests that Google had declared ‘Code Red’ and assigned additional people for its AI strategy.
With the co-founders getting involved, Google seems to be taking things very seriously indeed. After stepping away from their roles at the company back in 2019, Page and Brin had only checked in on the company occasionally, without interfering in its plans.
ChatGPT, Open AI’s artificial intelligence dialogue chatbot, has passed all three divisions of the United States Medical Licensing Examination, according to experts.
According to the latest research experiment, ChatGPT was comfortably within the passing range and depicted moderate accuracy in the exams. The name of the experiment is Performance of ChatGPT on USMLE: Potential for AI-Assisted Medical Education Using Large Language Models. The research is still being peer-reviewed.
The researchers wrote in the paper that ChatGPT performed at or almost near the passing threshold for all the three exams without any specialized training or reinforcement.
Researchers said the tool was capable of demonstrating a high level of insight and concordance in its explanation. They added that the results suggest that large language models may, in fact, have the potential to assist with medical education and clinical decision-making.
The first step of the exam was taken by medical students who have completed at least two years of learning. The second was taken by fourth-year medical students who have completed up to two years of clinical rotations. The step three was conducted by postgraduate students.
Google’s Parent Alphabet announced the layoff of 12000 of its employees for the year 2023. According to the reports, the layoff will be done globally and affect US staff immediately. The Chief Executive Officer, Sundar Pichai, has even mailed the employees saying he will take responsibility for the decision.
Besides Google, other big tech companies like Amazon and Microsoft announced layoffs at the beginning of the year. Along with tech companies, food delivery and grocery platforms like Zomato and Swiggy and online cab services like Ola have also announced significant layoffs.
Sundar Pichai has mentioned that he is confident of the huge opportunity in front of the organization and has thanked the strength of the company’s mission, the value of its products and services, and early investments in AI.
The layoff at Google Parent Alphabet will affect teams across the company like recruiting, corporate functions, product, and engineering teams. Alphabet has already emailed some affected employees, and the process can take longer in other countries due to local employment laws and practices.
According to an investigation by Time, OpenAI paid Kenyan workers less than $2 per hour to filter through tens of thousands of lines of text to help make ChatGPT safer to use.
The workers were assigned to label and filter out toxic data from ChatGPT’s training dataset. They were forced to read graphic details of NSFW content such as bestiality, murder, suicide, torture, self-harm, child sexual abuse, and incest, Time reported.
OpenAI partnered with a data labeling partner based in San Francisco, Sama, to detect and label toxic content that could further be fed as data into a filtering tool for ChatGPT. Sama claims to provide developing countries with “ethical” and “dignified digital work.”
Sama recruited data labelers in Kenya in order to work on behalf of OpenAI, thus playing a crucial role in making the chatbot safe for public usage.
Despite playing an integral role in building ChatGPT, the workers were faced with grueling conditions and low pay. The workers received wages between $1.32 and $2 an hour, based on seniority and performance.
Sama stopped its work for OpenAI in February 2022, about eight months earlier than the contracted period, because of the traumatic nature of the work and also because of an investigative report published by Time about Sama’s work with Meta.
In a significant setback, Google has lost its fight in Indian Supreme Court to block an antitrust order that will force the company to change the business model of its infamous Android operating system in a critical growth market.
In October, the Competition Commission of India (CCI) said that Google exploited its dominant position in Android and asked it to remove restrictions imposed on device makers, including those related to the pre-installation of apps. It also fined Google with $161 million.
Google challenged the order in court, saying it will hurt consumers and its business. The company warned growth of the Android ecosystem could stall and that it would be forced to change arrangements with more than 1,100 device manufacturers and about 1,000 app developers. Google also said that no other jurisdiction has ever asked for such far-reaching changes.
A three-judge bench at the Supreme Court, including India’s chief justice, delayed the implementation of the CCI’s directives by one week. However, they declined to block them. Chief Justice D.Y Chandrachud said, “We are not inclined to interfere.”
Google licenses its Android system to smartphone makers. However, critics say it imposes restrictions like mandatory pre-installation of its own apps that are seen as anti-competitive. Google argues that such agreements help keep Android free.
Why are so many professional people and college students surprised to learn a few of the basic facts about the field of data science? Maybe it’s because the discipline is relatively new, having grown out of statistics, computer programming, financial analysis, project management, and others. A decade ago, the term was unknown. Today, most top universities and colleges offer dozens of courses for students who want to get into the fast-growing career field.
What are the facts that surprise so many otherwise informed individuals? In addition to the fact that the best data science jobs go to degree-holders, people seldom know that collecting information is a core component of the process, that programming is not a prerequisite skill to enter the field, and that data science is not just a big tech profession. Consider the following details in order to get a clearer picture of what data science is all about.
The Best Jobs Go to Those with Degrees
There are a few myths surrounding education within the field of data science, so it’s important to parse them and deal with each one separately. First, there’s no need to hold a Ph.D. or even a grad school diploma to excel in the industry. However, getting a college degree is almost a requirement these days, particularly amid competition for the top jobs and career paths. For those contemplating data science as their future job category, step one is to figure out how to pay for college.
That’s why the great majority of prospective degree holders turn to private student loans as the most efficient method of covering school-related expenses for a four-year degree. The good news is that anyone can apply for a loan. Acquiring financing in advance is the smartest way to pay for college, especially when the goal is to learn real-world skills and secure a position as a data science expert upon graduation.
More Data is Not Always a Good Thing
Data science is about quality, not quantity. Gathering and cleaning bits and pieces of data-rich information can be a tedious process, but the resulting reports, charts, and other outputs are directly related to the quality of the input. The old programming acronym, GIGO, garbage in, garbage out, still applies and is of particular relevance to data scientists in the 2020s. Researchers can gather thousands of statistical studies, but if the original information that formed the basis of those reports was flawed, the results would not be relevant or usable. Instead of pushing for quantity, experienced data science professionals aim to locate as much high-quality information as possible. Quantity itself is not a bad thing as long as all the original input is useful and relevant.
It’s Not Just for Programmers
The data science field is not an exclusive club for computer programmers and math wizards. While there are a number of programming and math enthusiasts and experts who pursue data science as a career, most people are surprised to find a wide variety of other disciplines working in top jobs. One of the newest trends in business schools is for accounting, finance, marketing, and management majors to venture into data-based science positions after graduation.
Collecting Data is a Primary Part of the Science
So much emphasis in business literature focuses on cleaning, organizing, processing, and analyzing information, and it’s easy to forget about the collection phase. No matter what the industry is, professionals who gather statistics and other data spend a lot of time on two steps of the task. First, they need to identify where relevant bits of information might be hiding. Typical suspects include government reports, online public records, archived scientific files, hard-copy documents in libraries, and dozens of other locations. The next challenge is to gather those resources in the most efficient way possible.
