Connect with us

AI April 7, 2023

Meta Unveils AI Image Segmentation Model, SAM

Published

on

Meta Unveils AI Image Segmentation Model, SAM

Meta has unveiled its Segment Anything Model (SAM), an AI model that can identify individual objects within an image, including those it has not previously encountered.

Meta’s research division said it has published the new tool and corresponding dataset to foster research into foundation models for computer vision.

On its Twitter account, Meta AI wrote: “Today we’re releasing the Segment Anything Model (SAM) — a step toward the first foundation model for image segmentation.

“SAM is capable of one-click segmentation of any object from any photo or video + zero-shot transfer to other segmentation tasks.”

According to a company blog post, SAM was trained using the largest dataset of its kind using over 1 billion masks or objects on 11 million licensed images. Having access to the biggest dataset allows the AI model to segment images it has never encountered before.

“The model is designed and trained to be prompt, so it can transfer zero-shot to new image distributions and tasks,” said the company.

“We evaluate its capabilities on numerous tasks and find out that its zero-shot performance is impressive – often competitive with or even superior to prior fully supervised results.”

Also read: Google Claims its AI Computer Outperforms Nvidia’s A100 Chip

Meta is reportedly already using technology similar to SAM internally for tagging photos, content moderation and post suggestion on Facebook and Instagram.

The first foundation model for image segmentation?

The SAM model allows users to annotate images either by clicking on them or providing a text prompt. India.com posted an image with the word cat prompted, and SAM promptly drew boxes around cats within the image.

Responding to Meta’s tweet announcing the new tool, most users expressed enthusiasm. “Wow, this will accelerate the self-driving and robotics industry 10x,” responded Arkash, while another described it as “very cool.”

“It sounds like a significant step towards the development of the first foundation model for image segmentation. Keep up the fantastic work,” said IslandPitch.

But MagicOfBarca thinks otherwise, responding “Idk what’s the point of this or what are the main uses for it so hope you make a video on this.”

According to Meta, SAM uses an image encoder that produces a one-time embedding for the image, while a lightweight encoder converts any prompt into an embedding vector in real-time.

These two information sources are then combined in a lightweight decoder that predicts segmentation masks. After the image embedding is computed, SAM can produce a segment in just 50 milliseconds.

Feel like trying it?

SAM is developed by Meta AI Research and it is publicly available on GitHub. You can also try SAM online with a demo or download the dataset (SA-1B). Here’s what you need to do.

  1. Download the demo or go to the Segment Anything Model demo link.
  2. Upload an image or choose one in the gallery.
  3. Add and subject areas.
  4. Mask areas by adding points. Select Add Area, then select the object. Refine the mask by selecting.
  5. Remove Area, then select the area.

The AI arms race

Conversations around AI have exploded across the globe following the release of OpenAI’s ChatGPT in November. This has sparked serious competition in the sector, with tech giants like Meta, Microsoft and Google racing to come up with their own rival products or incorporate the tech in their products and services.

Microsoft has gone as far as adding ChatGPT’s capabilities to its Bing search engine, together with its Office suite.

Meta, meanwhile, has been experimenting with generative AI. CEO Mark Zuckerberg says incorporating such technology into the company’s apps is a priority this year, though it hasn’t abandoned the metaverse.

Among the AI tools the company is developing is one that creates surreal videos from text prompts. There’s also another that  quickly generates children’s book illustrations from prose.

SHARE THIS POST
Image credits: Shutterstock, CC images, Midjourney, Unsplash.

AI

Judge Orders All AI-Generated Research To Be Declared in Court

Published

on

Judge Orders All AI-Generated Research To Be Declared in Court

A Texas federal judge has ordered that AI-generated content should not be used to make arguments in court, and that such information must be declared and verified by a human.

Judge Brantley Starr’s ruling comes after one attorney, Steven Schwartz, last week allowed OpenAI’s ChatGPT to “supplement” his legal research by providing him with six cases and relevant precedent. All the cases were untrue and completely “hallucinated” by the chatbot.

Also read: ChatGPT’s Bogus Citations Land US Lawyer in Hot Water

The debacle received wide coverage, leaving Schwartz with “regrets.” Other lawyers who may have been contemplating trying the stunt now have to think twice, as Judge Starr has put an end to it.

Judge Starr also added a requirement that any attorney who appears in his courtroom declare that “no portion of the filing was drafted by generative artificial intelligence,” or if it was, that it was checked “by a human being.”

Judge Starr lays down the law

The eminent judge has set specific rules for his courtroom, just like other judges, and recently added the Mandatory Certification Regarding Generative Artificial Intelligence.

This states that: “All attorneys appearing before the Court must file on the docket a certificate attesting either that no portion of the filing was drafted by generative artificial intelligence (such as ChatGPT, Harvey.AI, or Google Bard) or that any language drafted by generative artificial intelligence was checked for accuracy, using print reporters or traditional legal databases, by a human being.”

A form for lawyers to sign is appended, noting that “quotations, citations, paraphrased assertions and legal analysis are all covered by this proscription.”

According to a report by TechCrunch, summary is one of AI’s strong suits and finding and summarizing precedent or previous cases is something advertised as potentially helpful in legal work. As such, this ruling may be a major spanner in the works for AI.

The certification requirement includes a pretty well-informed and convincing explanation of its necessity.

It states that: “These platforms are incredibly powerful and have many uses in the law: form divorces, discovery requests, suggested errors in documents, anticipated questions at oral argument.

“But legal briefing is not one of them. Here’s why.

“These platforms in their current states are prone to hallucinations and bias,” reads part of the certification.

It further explains that on hallucinations, AI is prone to simply making stuff up – even quotes and citations. While another issue relates to reliability or bias.

Chatbots don’t swear an oath

The certification further notes that although attorneys swear an oath to set aside their personal prejudices, biases, and beliefs to faithfully uphold the law and represent their clients, generative AI is the programming devised by humans who did not have to swear such an oath.

In the case of Schwartz, he said in an affidavit that he was “unaware of the possibility that its (ChatGPT) content could be false.”

He added that he “greatly regrets” using the generative AI and will only “supplement” its use with absolute caution and validation in future, further claiming he had never used ChatGPT prior to this case.

The other side of ChatGPT

Launched last November, ChatGPT is a large language model developed by OpenAI. The AI-powered chatbot is trained on billions of data sets from the internet and can perform a variety of tasks such as generating text and translating languages.

Despite going viral and provoking a fierce AI race, ChatGPT has its downsides – it can hallucinate and has misled Schwartz, who was representing Roberto Mata in a lawsuit against Colombian airline Avianca. Effectively, the chatbot provided citations to cases that did not exist.

Yet when Schwartz asked ChatGPT if one of the supposed cases was a real case, it responded “yes, (it) is a real case.” When asked for sources, the chatbot told Schwartz the case could be found “on legal research database such as Westlaw and LexisNexis.”

The matter came to light after the opposing counsel flagged the ChatGPT-generated citations as fake.

US District Court Judge Kevin Castel confirmed six of them as non-existent and demanded an explanation from Schwartz.

“Six of the submitted cases appear to be bogus judicial decisions with bogus quotes and bogus internal citations,” wrote Judge Castel in a May 4 order.

Continue Reading

AI

Nvidia Debuts AI Tools in an Era Where “Anyone Can Be a Programmer”

Published

on

Nvidia Debuts AI Tools in an Era Where “Anyone Can Be a Programmer”

The world’s most valuable chip maker Nvidia has unveiled a new batch of AI-centric products, as the company rides on the generative AI wave where anyone can be a programmer.

Nvidia announced a new supercomputer and a networking system, while the company also aims to make video game characters more realistic.

The wide range of products include robotics design, gaming capabilities, advertising services, and networking technology, which CEO Jensen Huang unveiled during a two-hour presentation in Taiwan on Monday.

Also read: Google Claims its AI Computer Outperforms Nvidia’s A100 Chip

Most notable of the new products is the AI supercomputer platform named DGX GH200 that will help tech companies create successors to OpenAI’s ChatGPT.

According to the company, the new DGX GH200 supercomputers combine 256 GH200 superchips that can act as a single graphics processing unit (GPU). The result is a system that boasts nearly 500 times the memory of a single Nvidia’s DGX A100 system.

“Generative AI, large language models, and recommender systems are the digital engines of modern economy,” said Huang.

“DGX GH200 AI supercomputers integrate Nvidia’s most advanced accelerated computing and networking technologies to expand the frontier of AI.”

So far, Microsoft Corp., Meta Platforms Inc., and Alphabet’s Google are expected to be among the first users, according to Nvidia.

The DGX GH200 supercomputers are expected to be available by the end of 2023.

The GH200 superchips which power the new supercomputer work by combining Nvidia’s Arm-based Grace GPU and an Nvidia H100 Tensor Core GPU in a single package.

The chipmaker also revealed that it’s building its own supercomputer running four DGX 200 systems at the same time to power its own research.

Nvidia also released its ACE generative AI model for video games, enabling gaming companies to use generative AI for large games with multiple non-player characters, giving them unique lines of dialogue and ways to interact with players that would normally need to be individually programmed.

Easy ad content

Alongside the hardware announcement, the company said it has partnered with advertising giant WPP to create a content engine that uses its Omniverse technology and generative AI capabilities to help build out ad content.

The move is intended to cut down the time and cost of producing ads by enabling WPP’s clients to lean on Nvidia’s technology.

Electronics manufacturers such as Foxconn, Pegatron, and Wistron are using Omniverse technology to create digital twins of their factory floors, so they can get a sense of how best to lay them out before making any physical changes.

A new computing era

Presenting at the forum, Huang acknowledged that advancements in AI are ushering in a new era in computing. He says anyone can be a programmer simply by speaking to the computer.

According to the Nvidia boss, gone are the days when programmers would write lines of code, only for it to display the “fail to compile” response because of a missing semicolon.

“This computer doesn’t care how you program it, it will try to understand what you mean, because it has this incredible large language model capability. And so the programming barrier is incredibly low,” said Huang.

“We have closed the digital divide. Everyone is a programmer. Now, you just have to say something to the computer,” he added.

Huang said his company has managed to bridge the digital gap, and the tech giant will continue to capitalize on the AI frenzy that has made Nvidia one of the world’s most valuable chipmakers.

Nvidia’s stock price is rising

Nvidia’s major announcements came as shares of the tech giant jumped last week on news that the company anticipated second quarter revenue above Wall Street’s expectations, based on the strength of its data center business.

The company hit the $1 trillion market cap just before the US markets opened on Tuesday. Its shares are trading at $407 in the pre-market, nearly 5% up from Monday.

Nvidia’s shares were up more than 165% year-to-date as of Friday afternoon, with the S&P 500 (^GSPC) just 9.5% higher in the same frame.

Rival chip maker AMD has experienced a similar boost in share price, rising 93%. However, Intel (INTC) is lagging behind with shares up just 8%.

According to Yahoo Finance tech editor Daniel Howley, while analysts see Nividia well ahead of its chip rivals in the AI processing space, how long that continues to be the case is anyone’s guess.

Continue Reading

AI

ChatGPT’s Bogus Citations Land US Lawyer in Hot Water

Published

on

ChatGPT's Bogus Citations Land US Lawyer in Hot Water

A lawyer in the United States is facing disciplinary action after his law firm used popular AI chatbot ChatGPT for legal research and cited fake cases in a lawsuit.

Steven A. Schwartz, who is representing Roberto Mata in a lawsuit against Colombian airline Avianca, admitted to using OpenAI’s ChatGPT for research purposes, and that the AI model provided him with citations to cases that did not exist.

Mata is suing Avianca for a personal injury caused by a serving cart in 2019, claiming negligence by an employee.

Also read: Opera Unveils GPT-Powered AI Chatbot Aria

Bogus all the way

According to a BBC report, the matter came to light after Schwartz, a lawyer with 30 years experience, used these cases as precedent to support Mata’s case.

But the opposing counsel flagged the ChatGPT-generated citations as fake. US District Court Judge Kevin Castel confirmed six of them as non-existent. He demanded an explanation from Schwartz, an attorney with New York-based law company Levidow, Levidow & Oberman.

“Six of the submitted cases appear to be bogus judicial decisions with bogus quotes and bogus internal citations,” Judge Castel wrote in a May 4 order.

“The court is presented with an unprecedented circumstance.”

The supposed cases include: Varghese v. China South Airlines, Martinez v. Delta Airlines, Shaboon v. EgyptAir, Petersen v. Iran Air, Miller v. United Airlines, and Estate of Durden v. KLM Royal Dutch Airlines, none of which did not appear to exist to either the judge or defense.

Lawyer claims ignorance

ChatGPT is a large language model developed by OpenAI. Launched in November, the AI is trained on billions of data from the Internet and can perform a variety of tasks like generate text, translate languages, and even write poetry, and solve difficult math problems.

But ChatGPT is prone to “hallucinations” – tech industry speak for when AI chatbots produce false or misleading information, often with confidence.

In an affidavit last week, Schwartz said he was “unaware of the possibility that its [ChatGPT] content could be false.” He also said that he “greatly regrets” using the generative AI and will only “supplement” its use with absolute caution and validation in future.

Schwartz claimed to have never used ChatGPT prior to this case. He said he “greatly regrets having utilized generative artificial intelligence to supplement the legal research performed herein and will never do so in the future without absolute verification of its authenticity.”

The career attorney now faces a court hearing on June 8 after accepting responsibility for not confirming the authenticity of the ChatGPT sources. Schwartz was asked to show cause why he shouldn’t be sanctioned “for the use of a false and fraudulent notarization.”

ChatGPT’s confident lies

According to the BBC report, Schwartz’s affidavit contained screenshots of the attorney that confirmed his chats with ChatGPT.

Schwartz asked the chatbot, “is varghese a real case?”, to which ChatGPT responded “yes, [it] is a real case.” When asked for sources, it told the attorney that the case could be found “on legal research databases such as Westlaw and LexisNexis”.

Again, the attorney asked: “Are the other cases you provided fake?” ChatGPT responded “No”, adding that the cases could be found on other legal databases. “I apologize for the confusion earlier,” ChatGPT said.

“Upon double-checking, I found the case Varghese v. China Southern Airlines Co. Ltd., 925 F.3d 1339 (11th Cir. 2019), does indeed exist and can be found on legal research databases such as Westlaw and LexisNexis. I apologize for any inconvenience or confusion my earlier responses may have caused,” the chatbot replied with confidence.

Continue Reading

News Feed

Advertise With Us

Unlock a wide range of advertising
opportunities with MetaNews to reach the
fast-paced web3 world.

Publish Your PR

Share your press release with
MetaNews’s growing global audience,
fans, and followers.

Subscribe for Email Updates

* indicates required

Copyright © 1997 – 2023 MetaNews All Rights Reserved

Copyright © 1997 - 2023 MetaNews All Rights Reserved

Welcome

Install
×