The iPhone could soon have Google’s Gemini artificial intelligence (AI) to power new features, following chats between Apple and the search giant.

According to a Bloomberg report, the two companies are reportedly in ‘active negotiations to let Apple license Gemini, coming to the iPhone software this year.’

Gemini is Google’s suite of generative AI tools, including chatbots and coding assistants. Over the years, Google is said to have paid Apple billions of dollars annually to make its search engine the default option in the Safari web browser on the iPhone and other devices.

A partnership between these two tech titans could give them a huge edge over the industry, but it seems more has to be done as the report says: “the two parties haven’t decided on the terms or branding of an AI agreement or finalized how it would be implemented.”

Apple has recently held discussions with OpenAI too, which is one of Gemini’s biggest competitors in the industry.

Neither Apple nor Google have confirmed or denied the potential for a collaboration.

Apple to focus on AI in the future

Coming in September 2024, Apple’s update – in the form of iOS 18 – is expected to be heavily focused on the use of AI.

Bloomberg’s Mark Gurman believes this could be a ‘relatively groundbreaking’ software update with ‘major new features and designs.’ Although the rumors have been circulating for the last few months, the Silicon Valley-based company hasn’t yet offered much in the way of sneak peeks.

This comes after news around Apple’s own testing of its large language model named Ajax. In 2023 we reported how the chatbot, built using Apple’s own framework named “Ajax,” is a large language model that utilizes AI to generate human-like responses. 

It’s believed that Ajax GPT was created for internal use and not much else has been heard about the tool over the last few months.

Featured Image: Via Ideogram

The post Apple is in talks to license Gemini AI for iPhones appeared first on ReadWrite.

Elon Musk has indicated his artificial intelligence (AI) startup xAI will open-source Grok, its chatbot rival to OpenAI’s ChatGPT.

The Tesla billionaire was once closely associated with OpenAI as a co-founder, but he walked away years ago to follow his own path in AI development, with a bitter fallout ensuing. Musk has taken legal action against his former company and its incumbent CEO, Sam Altman, for allegedly reneging on its original purpose to build AI for the benefit of humanity and not for the pursuit of profit.

In December 2023, Grok was rolled out to premium subscribers on X, the social media platform formerly known as Twitter, with the inclusion of a real-time feature exhibiting a thorough understanding of current events.

The AI assistant was described as “snarky and anti-woke” as part of a perceived agenda to challenge political correctness in digital spaces.

This week, @xAI will open source Grok

— Elon Musk (@elonmusk) March 11, 2024

How has Open AI responded to Elon Musk’s claims?

In recent days, OpenAI fired a sharp rebuke at Musk in response to his lawsuit, whilst dismissing the extent of his early involvement and influence within the most valuable AI startup.

An official blog post, penned collectively by Altman, Greg Brockman, John Schulman, Ilya Sutskever, and Wojciech Zaremba detailed how it had raised less than $45 million from Musk, despite his pledge to bring in up to $1 billion in funding.

OpenAI stated how it received more than $90 million from other donors to develop its research, seeking to place Musk on the fringes of its journey.

In response to his claims, the company implored it has remained true to its mission “to ensure AGI benefits all of humanity, which means both building safe and beneficial AGI”, “every step of the way”.

Another contentious aspect of the disagreement is the counter-claim Musk wanted OpenAI to merge with Tesla or to give him full control.

This is an intriguing, meaningful lawsuit in the overall landscape, which is likely to have a bearing on the balance of power in the AI race, as well as its direction.

Featured image: Midjourney

The post Elon Musk says xAI will open-source Grok amid OpenAI dispute appeared first on ReadWrite.

In a recent conversation between the CEOs of Microsoft and OpenAI, it was revealed by Sam Altman that ChatGPT-5 is expected to receive significant updates to its speech, images, and eventually video capabilities.

On his “Unconfuse Me” podcast, Bill Gates, along with Altman, explored the future of artificial intelligence, including its improved reasoning ability and general reliability. “Multimodality will be important,” Altman said, hinting at a future where artificial intelligence (AI) can perform increasingly complex tasks and potentially reshape various sectors, including programming, healthcare, and education.

Anticipation is building for the next iteration of ChatGPT, known as GPT-5. This advanced large language model is seen as a crucial milestone on the path to achieving artificial general intelligence (AGI), enabling machines to mimic human thought processes.

Here’s what to expect with the next version of GPT.

What is ChatGPT-5?

OpenAI says ChatGPT-5 “will be a state-of-the-art language model that makes it feel like you are communicating with a person rather than a machine.”

GPT-5 marks the next generation of the company’s Generative Pre-trained Transformer language model. OpenAI claim it represents a major advancement in natural language processing capabilities. With its more human-like ability to comprehend and produce text, GPT-5 could transform how we communicate with machines and automate numerous language-related jobs.

Will there be a ChatGPT-5 and what can it do?

As Altman has suggested, ChatGPT-5 is already in development as an updated version of its predecessor, GPT-4. The OpenAI CEO stated, “Right now, GPT-4 can reason only in extremely limited ways, and its reliability is also limited,” hence the aim is to improve its current functionality.

GPT, which stands for “Generative Pre-trained Transformer,” is a deep learning-based language model designed to produce text that resembles human writing. It boasts more natural language processing skills and finds widespread use across numerous applications.

On top of being dependable, Altman stipulated that “customizability and personalization will also be very important.

“People want very different things out of GPT-4; different styles, different sets of assumptions – we’ll make all that possible,” he added.

Altman highlighted that GPT-5’s ability to utilize personal data, including understanding emails, calendar details, appointment scheduling preferences, and integrating with external data sources, will be among the key advancements.

Multi-modal AI is designed to learn from and use a variety of content types such as images, audio, video, and numerical data. OpenAI has stated that GPT-4 is a multi-modal model, capable of processing both text and image inputs, although it is restricted to generating outputs in text form only, but GPT-5 would use more data to train on.

“We launched images and audio, and it had a much stronger response than we expected. We’ll be able to push that much further, but maybe the most important areas of progress will be around reasoning ability,” Altman told Gates on his podcast.

OpenAI has already indicated that it is working on a “supersmart” assistant to run a computer for its user. It is said to rival Microsoft and Google’s own AI workplace assistant but these programs are said to be in their infancy.

When will ChatGPT-5 be released?

However, Altman has not revealed a specific date for its release. He told the Financial Times in November that teams were working on the large language model, but did not state when this would be due.

Speaking at the World Governments Summit (WGS) in Dubai in February, Altman then reiterated that ChatGPT-5 is “going to be smarter.”

“It’s not like that this model is going to get a little bit better, it’s because we’re going to make them all smarter, it’s going to be better across the board,” he continued. He also spoke to Bloomberg, saying that he expected the company to “take its time” and make sure it can launch a product that they can feel “good about and responsible about.”

Despite the quick release of GPT-4 following ChatGPT, it underwent more than two years of training, development, and testing. Should GPT-5 follow a similar timeline, its arrival might be sometime in 2025. Nonetheless, this does not mean that we will not see any updates with GPT-4. OpenAI is expected to further develop GPT-4 and may even introduce an interim update, potentially labeled GPT-4.5, in the meantime.

Sam Altman addresses GPT-5 speculation during a recent Bloomberg interview:

“I don’t know what we’re going to call our next model…I don’t want to be like shipping iPhone 27…I expect us to take our time (developing)”#ChatGPT #ai #genai pic.twitter.com/BLMugmbhfp

— Equity Sesame (@EquitySesame) February 6, 2024

On at least two occasions last fall, Altman affirmed that OpenAI was actively developing GPT-5.

The initial confirmation came during a speech at the alumni reunion of Y Combinator, his former venture capital firm, last September, as corroborated by two attendees. At that event, Altman stated that GPT-5 and its successor GPT-6 “were in the bag,” implying their development was assured and that they would surpass the capabilities of previous models.

We have reached an agreement in principle for Sam Altman to return to OpenAI as CEO with a new initial board of Bret Taylor (Chair), Larry Summers, and Adam D'Angelo.

We are collaborating to figure out the details. Thank you so much for your patience through this.

— OpenAI (@OpenAI) November 22, 2023

Will ChatGPT-5 be free?

While there is a free version of ChatGPT, it is unclear whether ChatGPT-5 will require a subscription like its predecessor. The ChatGPT Plus subscription plan is $20 a month, providing subscribers with exclusive benefits including priority access during high-traffic periods, enhanced response times, the ability to use plugins, and exclusive access to GPT-4. Users also have access to its in-house AI image model DALL·E.

It’s also important to note that current language models are already expensive to train and maintain. This means that when GPT-5 is eventually released, access to it will likely require a subscription to ChatGPT Plus or Copilot Pro.

Ultimately, the launch of GPT-5 could lead to GPT-4 becoming more affordable and accessible. In the past, the high cost of GPT-4 has deterred a number of users. However, once it becomes cheaper and widely available, ChatGPT’s capability to handle complex tasks such as coding, translation, and research could significantly improve.

OpenAI has been approached for further comment.

Featured image: DALL·E / Canva

The post ChatGPT-5: release date, price, and what we know so far appeared first on ReadWrite.

When people refer to artificial intelligence (AI) tools, they may be referring more specifically to generative AI which is exploding in popularity and usage.

What is generative AI?

Generative AI is a term used to encompass deep-learning language models that generate text, images, videos, and other content using external prompts. They use the datasets they learn from to create various forms of content, ranging from short poems to long articles.

The topic of generative AI came to the fore with the wide release of OpenAI’s ChatGPT tool, bringing generative AI to the general public at large. While this might have been the first form of generative AI that many people had come into contact with, it certainly wasn’t the only one to hit the market.

There are now dozens of different ways that generative AI can be accessed, with almost every major tech company releasing its own versions, whether that’s Google’s Gemini, Microsoft’s massive investment in OpenAI, Github’s GitHub Copilot, and plenty more besides.

Before generative AI became synonymous with chatbots, it was also being used in statistics. It is an enormously helpful tool for analyzing numerical data, being able to quickly read through and identify patterns and trends. It was the growing strength of deep learning that made it possible to use the same technology for images, speech, and other complex data types.

The very first type of this new class of generative AI was introduced in 2013, known as variational autoencoders VAEs). These were widely used for generating realistic images and speech. Over time, the tools were fine-tuned to create ever more realistic images, sounds, and writing.

Introducing Sora, our text-to-video model.

Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. https://t.co/7j2JN27M3W

Prompt: “Beautiful, snowy… pic.twitter.com/ruTEWn87vf

— OpenAI (@OpenAI) February 15, 2024

How does generative AI work?

The definition of generative AI means that the tool is turning raw data (which can be anything from a quick prompt to scanning the entirety of a Wikipedia page) into output. The machine learning element means that the AI is learning what kind of output would be most likely, based on the data that it has access to. It cannot truly create anything new; genuine creativity still remains with living creatures alone.

However, it can create fresh pieces of output that are inspired by or similar to what it’s learned from, without it being a direct copy. Generally speaking, a prompt is given in any format, whether that’s text, an image, a video, a webpage, or any other input that the AI system has learned to process. It then uses its internal AI algorithms to create new content in response to the prompt, based off patterns it’s learned.

Digging into the deeper technology required, generative AI tools will usually combine several algorithms to process content. The input given will then be transformed into raw characters, such as turning paragraphs into basic letters, punctuation, and words or images into distinct visual elements. These basic characters are then encoded as vectors, something that the AI can use to create fresh output.

Bias in generative AI

It’s worth noting when explaining generative AI that this process of teaching language models can expose the possibility of teaching AI models human biases. If there is bias in the original data, whether unconscious or otherwise, that bias will show up in the output later created.

In real-world terms, this has resulted in AI creating images or text that have inaccurate identities or references in them. As just one example, Google recently had to shut down its image-generation tools within Gemini because of historical inaccuracies showing up in its results.

Is the United States a better place to live compared to Nazi Germany? Google Gemini says no. pic.twitter.com/kQTJMZW8a3

— The Rabbit Hole (@TheRabbitHole84) February 27, 2024

Generative AI criticism

This is not the only form of criticism leveled at generative AI. Artists of various different disciplines have complained that deep-learning models are ‘stealing’ from human-made art. After all, as established above, generative AI can’t learn from anything.

The data used to ‘teach’ AI models is usually sourced from artistic work. Legally, this must be artwork that is open to the public, but that hasn’t stopped people from resenting the fact that AI images can other forms of output can be created instantaneously, seemingly off the backs of human artists.

How to generate AI images

If you want to try out AI image generation for yourself, there are plenty of tools to choose from. Many graphic design tools, including Adobe, Pixlr, and Canva have incorporated generative AI into their offerings, but some dedicated tools get the job done as well. These include:

• Open AI’s DALL-E 2
• Imagine
• Midjourney
• Stable Diffusion

Of course, tools like these aren’t free, with all requiring a paid subscription in order to use them. Midjourney is included in ChatGPT’s $20 monthly subscription, DALL-E with the ChatGPT Plus plan for $20 per user, Stable Diffusion’s Basic plan starting at $27 per month, and Imagine starting from as low as $3 a month.

Once you have access to your tool of choice, you can start straight away. Most generative AI tools appear much like a chatbot, with a dialog box where you can type in your prompt. Getting the best out of your tool requires some nifty prompt use but it’s worth noting that the model will quite literally learn as it gets to know you.

You can ask it to tweak images as you go, remaking the same image in different styles or altering specific areas of the image. Every tool is different in the level of changes you can make, but most offer some sort of tool where you can pick out individual elements that can be altered in different iterations.

Featured image: Unsplash

The post What is generative AI? How does it work? appeared first on ReadWrite.

More than 200 of the world’s leading researchers in artificial intelligence (AI) have signed an open letter calling on big players in AI  like OpenAI, Meta, and Google to allow outside experts to independently evaluate and test the safety of their AI models and systems.

The letter argues that strict rules put in place by tech firms to prevent abuse or misuse of their AI tools are having the unintended consequence of stifling critical independent research aimed at auditing these systems for potential risks and vulnerabilities.

Prominent signatories include Stanford University’s Percy Liang, Pulitzer-winning journalist Julia Angwin, Renée DiResta from the Stanford Internet Observatory, AI ethics researcher Deb Raji, and former government advisor Suresh Venkatasubramanian.

What are the AI researchers concerned about?

The researchers say AI company policies that ban certain types of testing and prohibit violations of copyrights, generation of misleading content, or other abuses are being applied in an overly broad manner. This has created a “chilling effect” where auditors fear having their accounts banned or facing legal repercussions if they push the boundaries to stress-test AI models without explicit approval.

“Generative AI companies should avoid repeating the mistakes of social media platforms, many of which have effectively banned types of research aimed at holding them accountable,” the letter states.

The letter lands amid growing tensions, with AI firms like OpenAI claiming that The New York Times’ efforts to probe for copyright issues in ChatGPT amounted to “hacking.” Meta has updated terms threatening revocation if its latest language model is used to infringe intellectual property.

Researchers argue companies should provide a “safe harbor” allowing responsible auditing, as well as direct channels to responsibly report potential vulnerabilities found during testing, rather than having to resort to “gotcha” moments on social media.

“We have a broken oversight ecosystem,” said Borhane Blili-Hamelin of the AI Risk and Vulnerability Alliance. “Sure, people find problems. But the only channel to have an impact is these ‘gotcha’ moments where you have caught the company with its pants down.”

The letter and accompanying policy proposal aim to foster a more collaborative environment for external researchers to evaluate the safety and potential risks of AI systems impacting millions of consumers.

Featured image: Ideogram

The post 200 AI researchers urge OpenAI, Google, Meta to allow safety checks appeared first on ReadWrite.

After a lawsuit was announced last Friday (March. 1) from Elon Musk, the people behind OpenAI have shared what they claim to be historical email correspondence between them and the Tesla chief executive as they say they’re aiming to ‘dismiss all of Elon’s claims.’

Elon Musk is suing OpenAI, its CEO Sam Altman, co-founder Greg Brockman, as well as others, as he accuses them of violating contractual commitments established during the company’s inception when he helped to found the company in 2015.

The lawsuit contends the Microsoft-backed AI company’s shift towards profit-oriented objectives goes against the original agreement.

Today (Mar.6) OpenAI took to X to share their latest blog post highlighting information about their side of the story and their opinion on the suit.

In the release, they expand on three specific points. These include:

• We realized building AGI will require far more resources than we’d initially imagined

• We and Elon recognized a for-profit entity would be necessary to acquire those resources

• We advance our mission by building widely-available beneficial tools

The authors behind the blog, which includes CEO Sam Altman, go on to say: “We’re sad that it’s come to this with someone whom we’ve deeply admired – someone who inspired us to aim higher, then told us we would fail, started a competitor, and then sued us when we started making meaningful progress towards OpenAI’s mission without him.

“We are focused on advancing our mission and have a long way to go. As we continue to make our tools better and better, we are excited to deploy these systems so they empower every individual.”

In response, Elon Musk replied via X simply telling OpenAI to “change your name”. He also responded with a laughing face emoji to a post from user @realSharonZhou who said ‘more accurate to rename OpenAI to OpenEmail.’

Change your name

— Elon Musk (@elonmusk) March 6, 2024

The emails between OpenAI and Elon Musk

The email chain shared from OpenAI dates back to 2015, when the company was first founded. Elon Musk helped to cofound OpenAI, but later stepped away in 2018. OpenAI has published emails from the South African-born billionaire in an attempt to knock down his claims the company has strayed from its founding mission to serve as a non-profit check on the threats of generative AI.

In the first email which is said to be from the billionaire entrepreneur, it says ‘We need to go with a much bigger number than $100M to avoid sounding hopeless…’ The content in the email goes on to explain a ‘$1bn funding commitment’ may be more appropriate.

When looking at the next step for OpenAI, the details in the blog say ‘Elon wanted majority equity, initial board control, and to be CEO. In the middle of these discussions, he withheld funding.’

The blog post continues to read: ‘We couldn’t agree to terms on a for-profit with Elon because we felt it was against the mission…’

In an email shared by OpenAI, apparently from Elon Musk, he says OpenAI should ‘attach to Tesla as its cash cow.’

In one of the last emails to be made public, cofounder Ilya Sutskever explains further about the tool. In the blog, the authors say ‘Elon understood the mission did not imply open-sourcing AGI.’ They later shared that Elon allegedly responded with ‘Yup.’

Featured image: Midjourney

The post OpenAI hit back at Elon Musk lawsuit with leaked emails appeared first on ReadWrite.

OpenAI has announced that its ChatGPT program can now read aloud to its users. The function with the same name can recite the responses in one of five voice options. Read Aloud can be accessed through the web version of ChatGPT as well as the ChatGPT apps for iOS and Android.

ChatGPT can now read responses to you.

On iOS or Android, tap and hold the message and then tap “Read Aloud”. We’ve also started rolling on web – click the "Read Aloud" button below the message. pic.twitter.com/KevIkgAFbG

— OpenAI (@OpenAI) March 4, 2024

The new feature is available in 37 languages and can automatically detect the language of the text it is reading. So far, the option is available for GPT-4 and GPT-3.5 users.

OpenAI announced the feature in a post on X: “ChatGPT can now read responses to you. On iOS or Android, tap and hold the message and then tap “Read Aloud.” We’ve also started rolling on the web — click the “Read Aloud” button below the message.”

The company has been striving to enhance its multimodal capabilities, hence it introduced a voice chat feature in September 2023, which allows users to input prompts verbally without needing to type. However, this will allow people to have the chatbot responses read out aloud as well.

Users can press and hold the text to access the Read Aloud player on the mobile app, enabling them to play, pause, or rewind the narration. Meanwhile, the web version displays a speaker icon beneath the text.

The update comes as Anthropic launched its Claude 3 AI chatbot, which it claims has reached “near-human” levels in some tests. Many users took to X to respond to the announcement, saying it fell short of their expectations.

Yikes if this is your answer to Claude 3

— Wilson (@Wilson00000009) March 4, 2024

Read aloud on other AI platforms

In July 2023, Google incorporated the same feature into its Gemini chatbot — formerly known as Bard —  adding the option for it to vocalize its responses aloud. It supported 40 new languages, including  Arabic, Chinese, German, Hindi, and Spanish. 

Microsoft Edge’s Read Aloud feature now works in its Windows 11 Copilot chatbot. The feature is in the testing phase and is now available for everyone.

Featured image: Canva / DALL-E

The post OpenAI’s ChatGPT now has multilingual read out loud feature appeared first on ReadWrite.

Microsoft has filed a motion seeking to dismiss parts of The New York Times lawsuit against OpenAI, claiming it was “doomsday futurology.”

The New York Times initiated legal action against both companies, accusing them of using their articles for the training of ChatGPT’s large language models, as well as profiting from them.

In February, OpenAI requested a federal judge to dismiss parts of the same copyright lawsuit, accusing the newspaper of employing deceptive tactics to generate misleading evidence.

In the filing, lawyers representing the tech giant quoted the former head of the Motion Picture Association of America, Jack Valenti, who said, “The VCR is to the American film producer and the American public as the Boston Strangler is to the woman home alone.” They assert that the analogy represents “alarmism” against technology and note that the U.S. Supreme Court eventually rejected the motion on the same grounds.

Microsoft alluded that the VCR actually bolstered the entertainment sector by creating new avenues for revenue. Like OpenAI’s countersuit, the company elaborated on its stance regarding LLMs, describing them as a breakthrough in artificial intelligence.

Through its partnership with OpenAI, Microsoft said it aimed to “help bring their extraordinary power to the public,” because it “believes in LLMs’ capacity to improve the way people live and work.”

It also accused the NYT of using “might and its megaphone to challenge the latest profound technological advance: the Large Language Model.” It called for their case to be dismissed on three grounds including unfounded copyright infringement through GPT tools, and unproven Digital Millennium Copyright Act violations.

The Washington-based company also alleged that the news organization mischaracterized its copyright claims based on GPT-model outputs as state law misappropriation torts, suggesting that the GPT-based tools appropriate “time-sensitive” news and Wirecutter reviews.

What has The New York Times said?

Ian Crosby, a partner at Susman Godfrey and the lead attorney for The New York Times in this lawsuit, stated on Monday (March 6), “Microsoft doesn’t dispute that it worked with OpenAI to copy millions of The Times’s works without its permission to build its tools.

“Instead, it oddly compares L.L.M.s to the VCR even though VCR makers never argued that it was necessary to engage in massive copyright infringement to build their products,” he added.

Featured image: Canva

The post Microsoft seeks dismissal in New York Times lawsuit, cites ‘doomsday futurology’ appeared first on ReadWrite.

Amidst fierce competition from tech giants like OpenAI, Google, and Microsoft in the race to develop cutting-edge conversational artificial intelligence (AI), Anthropic’s launch of Claude 3 aims to disrupt the field.

The AI research company has launched its new chatbot which it claims has reached “near-human” levels in some tests, whilst measuring favorably against ChatGPT and Google Gemini across several industry benchmarks.

In an official blog released on Monday (Feb. 4). the details of the large language model (LLM) were disclosed striking a resemblance to Gemini with its own three model levels, Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus.

All “Claude 3 models can power live customer chats, auto-completions, and data extraction tasks where responses must be immediate and in real-time,” but they can be distinguished in terms of their power and functionality.

Haiku, the smallest version of Claude 3, was described as “the fastest and most cost-effective model on the market,” with an impressive output enabling analysis of a research paper with charts and graphs in less than three seconds.

Sonnet is the power behind the Claude.ai chatbot, available for free with an email sign-in, with Opus representing the largest-capacity LLM. It can be accessed with a $20 monthly subscription to the ‘Claude Pro’ service and it benefits from a multi-modal offering, going beyond previous versions to work with both text and image inputs.

Today, we're announcing Claude 3, our next generation of AI models.

The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision. pic.twitter.com/TqDuqNWDoM

— Anthropic (@AnthropicAI) March 4, 2024

How does Claude 3 compare to comparative models?

Anthropic, founded by former OpenAI personnel, has lauded the performance outcomes of Claude with Opus displaying higher graduate-level reasoning than GPT-4, scoring 14.7% better than its rival in tests involving maths, coding, and knowledge. Importantly, this latest generation LLM from the Amazon-backed company has exceeded its predecessors:

“For the vast majority of workloads, Sonnet is 2x faster than Claude 2 and Claude 2.1 with higher levels of intelligence. It excels at tasks demanding rapid responses, like knowledge retrieval or sales automation. Opus delivers similar speeds to Claude 2 and 2.1, but with much higher levels of intelligence.”

Anthropi stated Claude 3 has been trained using publicly available and internal data, with hardware from Amazon Web Services and Google Cloud technology.

Featured Image: Anthropic

The post Anthropic claim new Claude 3 AI chatbot outperforms ChatGPT and Gemini appeared first on ReadWrite.

Figure AI has reportedly been given a valuation of $2.6 billion from investors including Jeff Bezos, Nvidia, and Microsoft.

The startup working to build humanoid robots that can perform dangerous and undesirable jobs was able to raise $675 million, when it was seeking $500 million in a funding round led by Microsoft Corp.

Founded in 2022, the California-based company said that its priority will be in industries such as manufacturing, shipping and logistics, warehousing, and retail, “where labor shortages are the most severe.” Its founder Brett Adcock shared the news on X, stating that, “OpenAI & Figure signed a collaboration agreement to develop next-generation AI models for robots”

Excited to share: Figure raises $675M at $2.6B Valuation

+ OpenAI & Figure signed a collaboration agreement to develop next generation AI models for robots

Below are the details: pic.twitter.com/V57nn9P3oA

— Brett Adcock (@adcock_brett) February 29, 2024

“Our vision at Figure is to bring humanoid robots into commercial operations as soon as possible. This investment, combined with our partnership with OpenAI and Microsoft, ensures that we are well-prepared to bring embodied AI into the world to make a transformative impact on humanity,” said Adcock in a statement.

He added, “AI and robotics are the future, and I am grateful to have the support of investors and partners who believe in being at the forefront.”

Earlier this week, the company unveiled a video demonstrating Figure 01’s capabilities. This robot, connected by a tether, can walk on two legs and uses its fingers to lift a plastic crate. It then proceeds to walk a few additional steps before setting the crate down on a conveyor belt. The company already showcased a robot making coffee in January.

Last month we demonstrated Figure 01 making coffee only using neural networks

This is a fully learned, end-to-end visuomotor policy mapping onboard images to low level actions at 200hz

Next up: excited to push the boundaries on AI learning with OpenAIpic.twitter.com/DNAZWnaYK3

— Figure (@Figure_robot) February 29, 2024

The beginning of humanoid robots

Last year, the UN’s International Telecommunication Union held its first news conference featuring humanoid social robots. Nine robots, carefully crafted to resemble humans, took center stage at the news conference. One of the robots named Sophia was developed by Hanson Robotics, while others were programmed by Beyond Imagination, the University of Geneva, Hiroshi Ishiguro, Neura Robotics, Aidan Meller, SingularityNET, and Engineered Arts.

The event looked at networking to build new projects, calls to action, and partnerships. It featured talks from thought leaders as well as demonstrations of state-of-the-art AI solutions that could achieve global scale with the support of the international AI for Good community. 

1X Technologies also confirmed a breakthrough in teaching skills to its android EVE, thanks to special neural networks, which allow the bots to learn by watching videos of people doing activities. Like Figure AI, the Norway-based humanoid robotics firm is backed by OpenAI.

Featured image: Canva / DoD photo by Senior Master Sgt. Adrian Cadiz

The post Figure AI: Humanoid robot maker given $2.6bn valuation by Jeff Bezos appeared first on ReadWrite.

Elon Musk is suing OpenAI, its CEO Sam Altman, as well as others, accusing them of violating contractual commitments established during the company’s inception in 2015, when he helped to found the company in 2015.

The legal action initiated by the Tesla CEO also includes OpenAI’s co-founder Greg Brockman. The suit states that Brockman, alongside Altman, initially engaged Musk with the proposition of creating an open-source, nonprofit entity aimed at advancing artificial intelligence technology for the “benefit of humanity.” Musk co-founded OpenAI in 2015 but stepped down from the company’s board in 2018.

The lawsuit contends that the Microsoft-backed company’s shift towards profit-oriented objectives goes against the original agreement. It highlights that Musk believed that this pivot was an “existential threat in AGI.”

Lawyers representing Musk stated that “Mr. Altman approached Mr. Musk with a proposal: that they join forces to form a non-profit AI lab that would try to catch up to Google in the race for AGI, but it would be the opposite of Google.”

Together with Brockman, the lawsuit added that they reached a consensus on the foundational ethos of the new AI lab. According to this agreement, OpenAI “would be a nonprofit developing AGI for the benefit of humanity, not for a for-profit company seeking to maximize shareholder profits; and (b) would be open-source, balancing only countervailing safety considerations, and would not keep its technology closed and secret for proprietary commercial reasons.” 

The organization was reportedly created to stand as a counterbalance to Google’s DeepMind project. However, unlike its counterparts, OpenAI’s mission was to advance the cause of humanity, steering clear of the profit-driven motives that typically underpin private, for-profit entities.

Elon Musk competes with OpenAI

In January, the billionaire entrepreneur‘s artificial intelligence start-up xAI was in talks to “raise up to $6 billion.” xAI made headlines in 2023 when “Grok,” a chatbot positioned as an adversary to OpenAI‘s ChatGPT, was introduced.

Another company known as Groq trademarked its name during its fledgling stage in 2016, so it’s no surprise they sent a cease-and-desist letter to the Tesla billionaire over his decision to title his firm with such a similar name.

Featured image: Canva / U.S. Air Force / Trevor Cokley / TechCrunch

The post Elon Musk sues OpenAI and CEO Sam Altman over breach of contract appeared first on ReadWrite.

OpenAI has requested a federal judge to dismiss parts of a copyright lawsuit filed by The New York Times, accusing the newspaper of employing deceptive tactics to generate misleading evidence, according to a recent Reuters report. The lawsuit, which centers around the alleged unauthorized use of the Times’ copyrighted material to train OpenAI’s artificial intelligence systems, including the popular ChatGPT, has sparked a heated debate over the boundaries of copyright law and AI technology.

OpenAI’s defense, articulated in a recent filing in Manhattan federal court, argues that The New York Times contravened OpenAI’s terms of use by using “deceptive prompts” to force the AI to reproduce the newspaper’s content. OpenAI contends that this strategy was designed to create evidence for The New York Times’ lawsuit, undermining the integrity of the legal process. The filing criticizes the Times for not adhering to its own high journalistic standards, suggesting that the newspaper hired an external party to manipulate OpenAI’s products deliberately.

At the heart of this legal battle is the controversial question of whether AI’s training on copyrighted materials constitutes fair use — a principle that allows limited use of copyrighted material without permission for purposes such as news reporting, teaching, and research. Tech companies, including OpenAI, argue that their AI systems’ usage of copyrighted content is a fair use, essential for the development of AI technologies that could potentially shape a multitrillion-dollar industry. However, copyright owners, including The New York Times, contend that such practices infringe on their copyrights, unduly benefiting from their extensive investments in original content.

What is the case against OpenAI and Microsoft about?

The case against OpenAI and its primary financial supporter, Microsoft, is part of a broader trend of copyright lawsuits targeting tech companies over AI training practices. However, courts have yet to provide a clear verdict on the fair use question in the context of AI, with some infringement claims being dismissed due to insufficient evidence of AI-generated content resembling copyrighted works closely.

OpenAI’s filing emphasizes the challenges in using ChatGPT to systematically reproduce copyrighted articles, arguing that the instances cited by the Times were anomalies resulting from extensive manipulation. The company also posits that AI models acquiring knowledge from various sources, including copyrighted materials, is inevitable and cannot be legally prevented, drawing a parallel with traditional journalistic practices of re-reporting news.

As the lawsuit progresses, the outcome could have profound implications for the future of AI development and the application of copyright law in the digital age. A ruling in favor of OpenAI could solidify the legal standing of AI’s fair use of copyrighted materials, potentially accelerating the growth of AI technologies. Conversely, a decision favoring The New York Times could impose new limitations on how AI can be trained, impacting the evolution of AI capabilities and the tech industry’s trajectory.

The post OpenAI claims New York Times misused ChatGPT to fabricate lawsuit evidence appeared first on ReadWrite.

Mistral AI, a Paris-based AI startup, has announced its alternative to OpenAI and Anthropic with Mistral Large, its large language model. In a blog post, the company described Mistral Large as a “cutting-edge text generation model” with “top-tier reasoning capabilities.”

According to Mistral AI, can be used for “complex multilingual reasoning tasks” such as code generation, transformation, and reading comprehension.  It also launched its own answer to Chat GPT with Le Chat, which is currently only available in beta. Initially, Mistral AI emphasized its open-source focus as its main selling point. Its first model was released under an open-source license, but other, larger subsequent models have not.

What is Mistral Large?

Like OpenAI, Mistral AI offers Mistral Large via paid API and usage-based pricing. According to Tech Crunch, Mistral Large currently costs $24 per million output tokens and $8 per million of input tokens to query Mistral Large. Tokens, the outlet added, are designed to represent small chunks of words, usually divided into syllables. So, for instance, “ReadWrite” would be split into “read” and “write” and be separately processed by the AI language model.

Also, according to the outlet, Mistral AI does, by default, support context windows of 32,000 windows. This translates into over 20,000 English words and supports numerous other European languages like Italian, French, German, and Spanish.

But that’s not all. As mentioned, Mistral AI is launching Le Chat, it’s own version of Chat-GPT. It’s available at chat.mistral.ai and is currently a beta release.

Specifically, users can choose between three models: Mistral Large, Mistral Small, and Mistral Next — a prototype which, according to Tech Crunch, is “designed to be brief and concise.” For now, at least, Le Chat is free to use — but there’s a chance of this changing in the future.

Featured Image: Canva

The post France’s Mistral AI releases new model to rival GPT-4 appeared first on ReadWrite.

Deepfake videos generated by artificial intelligence (AI) could be mobilized by bad actors to influence the upcoming UK election a top government minister has warned.

Home Secretary James Cleverly said it could create a ‘perfect storm’ before Brits head to the polls, likely before the end of 2024.

Cleverly – in preparation for a meeting with Silicon Valley bosses – said that the time for deepfake videos disrupting elections across the world is well and truly here and ‘already in play’.

He went on to warn British voters opinions may be swayed if their social media platforms are filled with convincing AI-generated deepfakes. These forgeries could be made on behalf of UK adversaries, such as Russia and Iran. However, this possibility is not limited to the UK, with over half the global population set to take part in elections this year – such as the UK, US, and the EU.

Speaking to The Times, Cleverly said:

“The era of deepfake and AI-generated content to mislead and disrupt is already in play, increasingly today the battle of ideas and policies takes place in the ever-changing and expanding digital sphere.”

“The landscape it is inserted into needs its rules, transparency, and safeguards for its users. The questions asked about digital content and the sources of digital content are no less relevant than those asked about the content and sources at dispatch boxes, newsrooms, or billboard ads.”

UK action on deepfakes

Cleverly’s concern was also echoed by London Mayor Sadiq Khan who warned that deepfakes could be a big factor in a close election.

Khan has experience of the potency of deepfakes, as an AI-generated video of him making rousing remarks before Armistice Day (Britain’s version of Veterans Day) protests last year. He believes the regulation around deepfakes is not currently effective.

As reported in The Independent, Khan said:

“It’s not an issue of policing, it’s an issue of legislation and regulation.”

“My worry is, in a close election, a close referendum, these sorts of deepfake videos and audios can be the difference, but also my concern is, there are sometimes examples where these sorts of deepfake audios can lead to serious disturbances, particularly when emotions are running high,”

Tech giants rally against deepfakes

Earlier this month, the major players in generative AI; Meta, OpenAI, Microsoft and 17 others, agreed to work collaboratively to prevent deceptive AI content; however, this agreement failed to commit to banning deepfake videos.

Ahead of the upcoming European Parliament elections in June, Meta said it is setting up a dedicated to tackle the abuse of AI and the spread of disinformation, as concerns grow over the manipulative power of AI over voters.

In a blog post, Marco Pancini, Meta’s head of EU affairs said:

“As the election approaches, we’ll activate an Elections Operations Center to identify potential threats and put mitigations in place in real time,”

Meta also said it is working with 26 independent fact-checking organizations across the EU across 22 languages.

Featured image: Midjourney

The post UK election could be influenced by AI deepfakes appeared first on ReadWrite.

Match Group and OpenAI have signed a deal that will see the dating app giant receive over 1,000 ChatGPT enterprise licenses for its employees, as reported by TechCrunch.

The company behind apps such as Tinder, Match, OKCupid, Hinge, and PlentyOfFish explained its plans to leverage AI across a variety of its apps in its Q4 earnings report, to “enable our business to evolve their existing products and build disruptive new products and user experiences that meet the changing needs and expectations of their target audiences and drive user growth.”

This agreement appears to be the first step towards implementing those plans as the company moves to get its workforce to use and interact with the technology it hopes will improve user experience in 2024.

To celebrate the agreement and to stay on brand, the press release announcing the deal was written with the help of ChatGPT, and it even includes a quote from ChatGPT itself, which says: “I’m thrilled Match Group matched with me.

“Together, we’re not just breaking the ice; we’re melting it and reshaping the way work gets done.”

The release also claims ChatGPT promises to “be the wingman employees didn’t know they needed.”

How will Match Group’s workforce use ChatGPT?

The dating app giant says it will use ChatGPT-4 to aid with coding, design, analysis, building templates, and other daily tasks, including communications, hence the rather tongue-in-cheek press release.

However, the company has stressed that only trained and licensed employees will have access to OpenAI‘s technology to protect corporate data.

To be given access to these tools, employees will have to undergo training that focuses on the technology’s capabilities, limitations, and responsible use.

While there has been no confirmation of the cost of this agreement to Match Group, the company says it believes the use of OpenAI’s tools will make teams more productive.

Featured Image: Photo by Good Faces Agency on Unsplash

The post Match Group and OpenAI sign enterprise deal, ChatGPT writes the press release appeared first on ReadWrite.

An emerging artificial intelligence (AI) model is garnering a lot of attention thanks to its rapid response speed.

Groq, a California-based startup founded in 2016, has produced an impressive AI model which could quickly bring the company into competition with the likes of OpenAI’s Chat GPT.

The company uses LPU (language processing units) architecture instead of GPU (graphics processing unit), enabling more efficient and faster speeds. This is where it differs from traditional AI models which heavily rely on GPUs, which are both expensive and difficult to procure.

Wow, that's a lot of tweets tonight! FAQs responses.
• We're faster because we designed our chip & systems
• It's an LPU, Language Processing Unit (not a GPU)
• We use open-source models, but we don't train them
• We are increasing access capacity weekly, stay tuned pic.twitter.com/nFlFXETKUP

— Groq Inc (@GroqInc) February 19, 2024

Groq’s offering is returning incredible efficiency with its breakneck speed.

Early indications appear to show Groq’s hardware outperforming ChatGPT with public benchmarks hitting blazing speeds of 500 tokens per second, compared to 30-50 for GPT 3.5.

This side-by-side demo of Groq vs. GPT 3.5 shows Groq completing the same prompt but ~4x faster.

The first public demo using Groq: a lightning-fast AI Answers Engine.

It writes factual, cited answers with hundreds of words in less than a second.

More than 3/4 of the time is spent searching, not generating!

The LLM runs in a fraction of a second.https://t.co/dVUPyh3XGV https://t.co/mNV78XkoVB pic.twitter.com/QaDXixgSzp

— Matt Shumer (@mattshumer_) February 19, 2024

What’s the difference between Groq and Grok?

If you are unsure of the name of the company and think you might have heard of it before, we need to mention Elon Musk.

After he distanced himself from OpenAI to concentrate on plans of his own, he announced the arrival of Grok late last year, bringing him into direct contention with OpenAI’s ChatGPT.

Groq trademarked its name during its fledgling stage in 2016, so it’s no surprise they sent a cease-and-desist letter to the Tesla billionaire over his decision to title his firm with such a similar name.

The irreverent statement brimmed with self-confidence but it did not miss its target with a stern underlying tone addressing Musk directly, as reflected in its intro:

“Did you know that when you announced the new xAI chatbot, you used our name? Your chatbot is called Grok and our company is called Groq®, so you can see how this might create confusion among people. Groq (us) sounds a lot like (identical) to Grok (you), and the difference of one consonant (q, k) only matters to scrabblers and spell checkers. Plus, we own the trademark.”

This won’t be the last we hear of this confrontation, stay tuned.

Image: Groq Inc/X.

The post Groq creates fastest generative AI in the world, blazing past ChatGpt and Elon Musk’s Grok appeared first on ReadWrite.

OpenAI is introducing a significant update to ChatGPT, integrating a “memory” feature that enables the AI to retain information about users over time, according to a recent report by The Verge. This enhancement aims to personalize interactions, allowing ChatGPT to recall specific details about users’ preferences, habits, and previous conversations.

The memory function operates in two primary ways: users can directly inform ChatGPT of certain details to remember, such as coding preferences, names of colleagues, or personal likes and dislikes. Alternatively, ChatGPT can autonomously learn and store information based on ongoing interactions with the user. This capability is designed to make the AI more intuitive and user-friendly, reducing the need for repetitive instructions.

Enhanced memory personalization across ChatGPT versions

Each version of GPT, including specialized ones like Books GPT, will possess its own memory, enhancing its ability to provide tailored recommendations and responses. For instance, a Books GPT with memory activated could remember a user’s reading history and genre preferences, offering more relevant suggestions. Similarly, educational, travel, and fitness-related GPT applications could benefit from remembering user-specific information to improve service delivery.

However, the introduction of memory to ChatGPT raises significant privacy and data security concerns. OpenAI’s method of learning from user interactions and developing a profile over time is reminiscent of other internet services’ data collection practices, which have often been met with skepticism and discomfort from users. The concern is that ChatGPT, by “knowing” users, might infringe on personal privacy.

To address these concerns, OpenAI emphasizes user control over ChatGPT’s memory. Users can inquire about what information ChatGPT retains about them and have the option to delete specific memories or manage them through a dedicated settings section. OpenAI also offers a Temporary Chat feature, akin to an incognito mode, allowing users to interact with ChatGPT without influencing its memory. Users can also disable the memory feature entirely across their accounts.

By default, the memory feature will be activated, and the data collected will contribute to training OpenAI’s models. However, data from ChatGPT Enterprise and Teams users will not be used in model training.

The post OpenAI introduces memory to ChatGPT for personalized conversations appeared first on ReadWrite.

OpenAI staff faced a protest as they were leaving work on Monday evening, with two groups calling on the workers to halt their work on advanced artificial intelligence (AI) systems like ChatGPT.

The groups known as Pause AI and No AGI are concerned about the rapid advances of AI and the danger that its capacity could overtake human control as well as cooperation with military authorities.

Join us and tell OpenAI "Stop working with the Pentagon!"
On 2/12, we will demand that OpenAI end its relationship with the Pentagon and not take any military clients. If their ethical and safety boundaries can be revised out of convenience, they cannot be trusted. pic.twitter.com/29w0c8LgG5

— Holly Elmore (@ilex_ulmus) February 6, 2024

Dozens gathered at the OpenAI premises on Bryant Street in San Francisco in the aftermath of the company amending its usage policy to omit details of using AI for military purposes reports Venture Beat. Days later, it was confirmed OpenAI had entered into a partnership with the Pentagon to work on “cybersecurity tools“.

The protest organizers spoke about their aims and why they were taking this action.

“The goal for No AGI is to spread awareness that we really shouldn’t be building AGI (artificial general intelligence) in the first place,” said its leader, Sam Kirchener.

“Instead we should be looking at things like whole brain emulation that keeps human thought at the forefront of intelligence.”

His counterpart at Pause AI, lead organizer Holly Elmore added their wish for “a global, indefinite pause on frontier development of AGI until it’s safe.”

She opined, “I would be so happy if they ended their relationship with the military. That seems like a really important boundary.”

The cause of OpenAI protests

OpenAI’s foray into the military space is a significant development, so it would be remarkable if they were to reverse their decision and grant Elmore her wish.

It comes at the same time the company’s CEO Sam Altman has warned of the potential dangers of AI due to “societal misalignments” which some would argue include the abuse of the technology for military means.

Altman has also called for the creation of a universal regulator for AI, a sentiment shared by Pause AI’s Elmore who added, “Self-governance is not enough for these companies, there really needs to be external regulation.”

Kirchener was blunt in his warning of where all this would lead if the advances of technology were left unchecked, calling for human dynamics to be prioritized.

“If we build AGI, there’s the risk that in a post-AGI world, we’ll lose a lot of meaning from what’s called the psychological threat, where AGI does everything for everyone. People won’t need jobs. And in our current society, people derive a lot of meaning from their work, he implored.

As reflected in their names, Pause AI appears more concerned with the safe development of new tech whilst No AGI wants it to be stopped altogether.

Image: Pexels/Andrew Neel

The post ‘Stop working with Pentagon’ – OpenAI staff face protests appeared first on ReadWrite.

OpenAI CEO Sam Altman has made guarded comments on the importance of mitigating the dangers of artificial intelligence (AI), as he revealed concerns that keep him awake at night.

He was speaking at the World Government Summit in Dubai, UAE via video link when he called for a universal regulator to play a crucial role in safeguarding the future advances of AI, to stave off the threat of serious damage to society. The 38-year-old leader referenced the International Atomic Energy Agency as the type of body he envisages for this industry including a distanced relationship with developers, as reported by ABC News.

“There’s some things in there that are easy to imagine where things really go wrong and I’m not that interested in the killer robots walking on the street direction of things going wrong,” said Altman.

“I’m much more interested in the very subtle societal misalignments where we just have these systems out in society and through no particular ill intention, things just go horribly wrong,” he added.

Time for action

Striking the right tone with his audience, Altman indicated companies like OpenAI – the maker of ChatGPT – should be bound by an organization in an oversight role with input from many players in the tech space globally.

“We’re still in the stage of a lot of discussion. So there’s you know, everybody in the world is having a conference. Everyone’s got an idea, a policy paper, and that’s OK,” Altman continued.

“I think we’re still at a time where debate is needed and healthy, but at some point in the next few years, I think we have to move towards an action plan with real buy-in around the world.”

Following his return to OpenAI in December, the boss of the Microsoft-backed company spoke of societal apprehensions about advanced AI, when announcing plans for an AI ethics council. Hence, his comments at the World Government Summit reflect a consistent approach.

Image: Midjourney

The post OpenAI CEO Sam Altman reveals AI danger which keeps him awake at night appeared first on ReadWrite.

The CEO of OpenAI, Sam Altman, is seeking a $7 trillion investment to overhaul the world’s semiconductor industry, reports The Wall Street Journal.

Semiconductor chips are a hugely limiting factor in the world of artificial intelligence (AI) development due to the extremely high computing power needed to train AI models. With this investment, Altman is looking to negate a factor that significantly limits OpenAI’s growth potential.

Despite huge investment and strategic focus from the US government, the semiconductor chip industry is complex. Due to its strategic value and the importance of chips to AI development, it is a highly regulated industry and the US has tight controls on the export of chips. Altman has met with Commerce Secretary Gina Raimondo to discuss his ambitious plan.

“OpenAI has had productive discussions about increasing global infrastructure and supply chains for chips, energy, and data centers—which are crucial for AI and other industries that rely on them,” said a spokesperson for OpenAI to the WSJ. “We will continue to keep the U.S. government informed given the importance to national priorities, and look forward to sharing more details at a later date.”

However, fundraising the $5 – 7 trillion dollars required will be no easy feat. As the WSJ reports, the size of the global semiconductor chip market was $527 billion last year and isn’t expected to reach $1 trillion until 2030. The investment sought dwarfs even the market for chip manufacturing equipment, which was approximately $100 billion last year.

Sam Altman is looking for an amount greater than most country’s GDP

Let’s just stop for a moment and put this figure in context because $7 trillion is an extravagantly large amount of money. The GDP of the United Kingdom is $3.1 trillion, less than half of what Alman is seeking. The combined value of Microsoft and Apple, the USA’s highest-valued businesses, is just over $6 trillion.

The tech entrepreneur has met with representatives from the government of the United Arab Emirates and businesspeople from Taiwan to try and advance his plans. Despite Middle Eastern interest, Altman wants to focus his chip plants in the US, in part due to the huge investment coming from the Biden administration into the chip industry.

OpenAI is partnered with Microsoft and has their full support. Altman has discussed his plans with Satya Nadella, Microsoft CEO, and Kevin Scott, its CTO, and is proceeding with their buy-in.

Nothing has yet been confirmed regarding this enormous plan, but Altman continues to meet with powerful people to push it forward.

Featured image credit: Midjourney

The post OpenAI founder Sam Altman is ‘seeking $7 trillion investment’ appeared first on ReadWrite.