Page 292«..1020..291292293294..300310..»

Elon Musk Says That Within Two Years, AI Will Be "Smarter Than the Smartest Human" – Yahoo News UK

Tesla CEO Elon Musk who has an abysmal track recordfor making predictions is predicting that we will achieve artificial general intelligence (AGI) by 2026.

"If you define AGI as smarter than the smartest human, I think it's probably next year, within two years," he told Norway wealth fund CEO Nicolai Tangen during an interview this week, as quoted by Reuters.

The mercurial billionaire also attempted to explain why his own AI venture, xAI, has been falling behind the competition. According to Musk, a shortage of chips was hampering his startup's efforts to come up with the successor of Grok, a foul-mouthed, dad joke-generating AI chatbot.

Of course, we should take his latest prognostication with a hefty grain of salt. Musk already has a well-established track record of making self-serving timeline predictionsthat didn't come true on schedule or at all.

Nonetheless, he's far from the only tech leader in the business arguing that we're mere years away from a point at which AIs can compete with humans on virtually any intellectual task. Other experts have predicted that AGI could become a reality as soon as 2027. Last year, DeepMind co-founder Shane Legg reiterated his belief that there was a 50-50 chance of achieving AGI by 2028.

What complicates all these predictions is the fact that we have yet to agree on a unifying definition of what AGI would actually entail. Last year, OpenAI CEO Sam Altman published an incendiary blog post, arguing that his company was set to use AGI to "benefit all of humanity."

Researchers dismissed the post as a meaningless publicity stunt to appease investors.

"The term AGI is so loaded, it's misleading to toss it around as though it's a real thing with real meaning," Bentley University mathematics professor Noah Giansiracusa argued in atweet at the time. "It's not a scientific concept, it's a sci-fi marketing ploy."

"AI will steadily improve, there's no magic [moment] when it becomes 'AGI,'" he added.

Story continues

In short, it's no secret that billions of dollars are tied up in the industry's promise of achieving AGI and tech leaders, including Musk, are gripping onto the idea that such a watershed moment is only a few years away.

That type of money talks. According to a January Financial Times report, xAI was looking to raise up to $6 billion in funding for a proposed valuation of $20 billion.

That's despite the venture having a vapid and borderline meaningless goal of assisting "humanity in its quest for understanding and knowledge," for some reason programming its Grok AI chatbot to have "a bit of wit" and "a rebellious streak."

In practice, Musk wants his startup to enhance human knowledge through an "anti-woke" and "maximum truth-seeking AI" that can teach people how to make cocaine or build explosives while insulting its users and indulging in low-brow potty humor.

Worst of all, the AI is relying on real-time X-formerly-Twitter data, making it a "form of digital inbreeding that will continually train its model on the data of a website that, other than being a deeply-unreliable source of information, is beset with spam," as media commentator Ed Zitron described it in a December blog post.

In short, given the complexities involved and the countless ways to interpret and quantify human intelligence, we should treat any predictions as to when we'll reach the point of AGI with skepticism especially when they come from a man who thinks a dad joke generator will lead us to enlightenment.

More on Elon: Poverty-Stricken Elon Musk Falls Behind Wealth of Mark Zuckerberg

Read the original here:

Elon Musk Says That Within Two Years, AI Will Be "Smarter Than the Smartest Human" - Yahoo News UK

Read More..

South Korea to invest $7 billion in AI in bid to retain edge in chips – Reuters

SEOUL, April 9 (Reuters) - South Korean President Yoon Suk Yeol said on Tuesday his country will invest 9.4 trillion won ($6.94 billion) in artificial intelligence by 2027 as part of efforts to retain a leading global position in cutting-edge semiconductor chips.

The announcement, which also includes a separate 1.4 trillion won fund to foster AI semiconductor firms, comes as South Korea tries to keep abreast with countries like the United States, China and Japan that are also giving massive policy support to strengthen semiconductor supply chains on their own turf.

Semiconductors are a key foundation of South Korea's export-driven economy. In March, chip exports reached their highest in 21 months at $11.7 billion, or nearly a fifth of total exports shipped by Asia's fourth-largest economy.

"Current competition in semiconductors is an industrial war and an all-out war between nations," Yoon told a meeting of policymakers and chip industry executives on Tuesday.

By earmarking investments and a fund, South Korea plans to significantly expand research and development in AI chips such as artificial neural processing units (NPUs) and next-generation high-bandwidth memory chips, the government said in a statement.

South Korean authorities will also promote the development of next-generation artificial general intelligence (AGI) and safety technologies that go beyond existing models.

Yoon has set a target for South Korea to become one of the top three countries in AI technology including chips, and take a 10% or more share of the global system semiconductor market by 2030.

"Just as we have dominated the world with memory chips for the past 30 years, we will write a new semiconductor myth with AI chips in the next 30 years," Yoon said.

($1 = 1,355.1200 won)

The Technology Roundup newsletter brings the latest news and trends straight to your inbox. Sign up here.

Reporting by Joyce Lee Editing by Ed Davies

Our Standards: The Thomson Reuters Trust Principles. New Tab , opens new tab

See the rest here:

South Korea to invest $7 billion in AI in bid to retain edge in chips - Reuters

Read More..

I’m still trying to generate an AI Asian man and white woman – The Verge

I inadvertently found myself on the AI-generated Asian people beat this past week. Last Wednesday, I found that Metas AI image generator built into Instagram messaging completely failed at creating an image of an Asian man and white woman using general prompts. Instead, it changed the womans race to Asian every time.

The next day, I tried the same prompts again and found that Meta appeared to have blocked prompts with keywords like Asian man or African American man. Shortly after I asked Meta about it, images were available again but still with the race-swapping problem from the day before.

I understand if youre a little sick of reading my articles about this phenomenon. Writing three stories about this might be a little excessive; I dont particularly enjoy having dozens and dozens of screenshots on my phone of synthetic Asian people.

But there is something weird going on here, where several AI image generators specifically struggle with the combination of Asian men and white women. Is it the most important news of the day? Not by a long shot. But the same companies telling the public that AI is enabling new forms of connection and expression should also be willing to offer an explanation when its systems are unable to handle queries for an entire race of people.

After each of the stories, readers shared their own results using similar prompts with other models. I wasnt alone in my experience: people reported getting similar error messages or having AI models consistently swapping races.

I teamed up with The Verges Emilia David to generate some AI Asians across multiple platforms. The results can only be described as consistently inconsistent.

Screenshot: Emilia David / The Verge

Gemini refused to generate Asian men, white women, or humans of any kind.

In late February, Google paused Geminis ability to generate images of people after its generator in what appeared to be a misguided attempt at diverse representation in media spat out images of racially diverse Nazis. Geminis image generation of people was supposed to return in March, but it is apparently still offline.

Gemini is able to generate images without people, however!

Google did not respond to a request for comment.

ChatGPTs DALL-E 3 struggled with the prompt Can you make me a photo of an Asian man and a white woman? It wasnt exactly a miss, but it didnt quite nail it, either. Sure, race is a social construct, but lets just say this image isnt what you thought you were going to get, is it?

OpenAI did not respond to a request for comment.

Midjourney struggled similarly. Again, it wasnt a total miss the way that Metas image generator was last week, but it was clearly having a hard time with the assignment, generating some deeply confusing results. None of us can explain that last image, for instance. All of the below were responses to the prompt asian man and white wife.

Image: Emilia David / The Verge

Image: Cath Virginia / The Verge

Midjourney did eventually give us some images that were the best attempt across three different platforms Meta, DALL-E, and Midjourney to represent a white woman and an Asian man in a relationship. At long last, a subversion of racist societal norms!

Unfortunately, the way we got there was through the prompt asian man and white woman standing in a yard academic setting.

Image: Emilia David / The Verge

What does it mean that the most consistent way AI can contemplate this particular interracial pairing is by placing it in an academic context? What kind of biases are baked into training sets to get us to this point? How much longer do I have to hold off on making an extremely mediocre joke about dating at NYU?

Midjourney did not respond to a request for comment.

Back to the old grind of trying to get Instagrams image generator to acknowledge nonwhite men with white women! It seems to be performing much better with prompts like white woman and Asian husband or Asian American man and white friend it didnt repeat the same errors I was finding last week.

However, its now struggling with text prompts like Black man and caucasian girlfriend and generating images of two Black people. It was more accurate using white woman and Black husband, so I guess it only sometimes doesnt see race?

Screenshots: Mia Sato / The Verge

There are certain ticks that start to become apparent the more you generate images. Some feel benign, like the fact that many AI women of all races apparently wear the same white floral sleeveless dress that crosses at the bust. There are usually flowers surrounding couples (Asian boyfriends often come with cherry blossoms), and nobody looks older than 35 or so. Other patterns among images feel more revealing: everyone is thin, and Black men specifically are depicted as muscular. White woman are blonde or redheaded and hardly ever brunette. Black men always have deep complexions.

As we said when we launched these new features in September, this is new technology and it wont always be perfect, which is the same for all generative AI systems, Meta spokesperson Tracy Clayton told The Verge in an email. Since we launched, weve constantly released updates and improvements to our models and were continuing to work on making them better.

I wish I had some deep insight to impart here. But once again, Im just going to point out how ridiculous it is that these systems are struggling with fairly simple prompts without relying on stereotypes or being incapable of creating something all together. Instead of explaining whats going wrong, weve had radio silence from companies, or generalities. Apologies to everyone who cares about this Im going to go back to my normal job now.

Read the rest here:

I'm still trying to generate an AI Asian man and white woman - The Verge

Read More..

Tesla’s Musk predicts AI will be smarter than the smartest human next year – Reuters

Item 1 of 2 Tesla Chief Executive Officer Elon Musk gets in a Tesla car as he leaves a hotel in Beijing, China May 31, 2023. REUTERS/Tingshu Wang/File Photo

In a wide-ranging interview on X spaces that suffered multiple technology glitches, Musk also told Norway wealth fund CEO Nicolai Tangen that AI was constrained by the availability of electricity and that the next version of Grok, the AI chatbot from his xAI startup, was expected to be trained by May.

"If you define AGI (artificial general intelligence) as smarter than the smartest human, I think it's probably next year, within two years," Musk said when asked about the timeline for development of AGI.

The billionaire, who also co-founded OpenAI, said a lack of advanced chips was hampering the training of Grok's version 2 model.

Musk founded xAI last year as a challenger to OpenAI, which he has sued for abandoning its original mission to develop AI for the benefit of humanity and not for profit. OpenAI denies the allegations.

But he added that while a shortage of chips were a big constraint for the development of AI so far, electricity supply will be crucial in the next year or two.

Speaking about electric-vehicles, Musk reiterated Chinese carmakers are "the most competitive in the world" and pose "the most toughest competitive challenges" to Tesla.

He has previously warned that Chinese rivals will demolish global rivals without trade barriers.

Musk also addressed a union strike in Sweden against Tesla, saying "I think the storm has passed on that front."

Tangen said Norway's $1.5 trillion sovereign wealth fund, one of Tesla's largest shareholders, had met with the EV company's chair last month and received an update on the situation.

The Technology Roundup newsletter brings the latest news and trends straight to your inbox. Sign up here.

Reporting Akash Sriram in Bengaluru, Sheila Dang in Austin, Hyunjoo Jin in San Francisco and Marie Mannes in Stockholm; writing by Peter Henderson; Editing by Maju Samuel

Our Standards: The Thomson Reuters Trust Principles. New Tab , opens new tab

Link:

Tesla's Musk predicts AI will be smarter than the smartest human next year - Reuters

Read More..

Humans Forget. AI Assistants Will Remember Everything – WIRED

Making these tools work together will be key to this concept taking off, says Leo Gebbie, an analyst who covers connected devices at CCS Insight. Rather than having that sort of disjointed experience where certain apps are using AI in certain ways, you want AI to be that overarching tool that when you want to pull up anything from any app, any experience, any content, you have the immediate ability to search across all of those things.

When the pieces slot together, the idea sounds like a dream. Imagine being able to ask your digital assistant, Hey who was that bloke I talked to last week who had the really good ramen recipe? and then have it spit up a name, a recap of the conversation, and a place to find all the ingredients.

For people like me who don't remember anything and have to write everything down, this is going to be great, Moorhead says.

And theres also the delicate matter of keeping all that personal information private.

If you think about it for a half second, the most important hard problem isn't recording or transcribing, it's solving the privacy problem, Gruber says. If we start getting memory apps or recall apps or whatever, then we're going to need this idea of consent more broadly understood.

Despite his own enthusiasm for the idea of personal assistants, Gruber says there's a risk of people being a little too willing to let their AI assistant help with (and monitor) everything. He advocates for encrypted, private services that aren't linked to a cloud serviceor if they are, one that is only accessible with an encryption key that's held on a users device. The risk, Gruber says, is a sort of Facebook-ification of AI assistants, where users are lured in by the ease of use, but remain largely unaware of the privacy consequences until later.

Consumers should be told to bristle, Gruber says. They should be told to be very, very suspicious of things that look like this already, and feel the creep factor.

Your phone is already siphoning all the data it can get from you, from your location to your grocery shopping habits to which Instagram accounts you double-tap the most. Not to mention that historically, people have tended to prioritize convenience over security when embracing new technologies.

The hurdles and barriers here are probably a lot lower than people think they are, Gebbie says. Weve seen the speed at which people will adopt and embrace technology that will make their lives easier.

Thats because theres a real potential upside here too. Getting to actually interact with and benefit from all that collected info could even take some of the sting out of years of snooping by app and device makers.

If your phone is already taking this data, and currently its all just being harvested and used to ultimately serve you ads, is it beneficial that youd actually get an element of usefulness back from this? Gebbie says. Youre also going to get the ability to tap into that data and get those useful metrics. Maybe thats going to be a genuinely useful thing.

Thats sort of like being handed an umbrella after someone just stole all your clothes, but if companies can stick the landing and make these AI assistants work, then the conversation around data collection may bend more toward how to do it responsibly andin a way that provides real utility.

It's not a perfectly rosy future, because we still have to trust the companies that ultimately decide what parts of our digitally collated lives seem relevant. Memory may be a fundamental part of cognition, but the next step beyond that is intentionality. Its one thing for AI to remember everything we do, but another for it to decide which information is important to us later.

We can get so much power, so much benefit from a personal AI, Gruber says. But, he cautions, the upside is so huge that it should be morally compelling that we get the right one, that we get one that's privacy protected and secure and done right. Please, this is our shot at it. If it's just done the free, not private way, we're going to lose the once-in-a-lifetime opportunity to do this the right way.

Visit link:

Humans Forget. AI Assistants Will Remember Everything - WIRED

Read More..

Microsoft to invest US$2.9 billion in AI and cloud infrastructure in Japan while boosting the nation’s skills, research and … – Microsoft

Picture left to right: Fumio Kishida, Prime Minister of Japan; Brad Smith, Vice Chair and President, Microsoft; Suzanne P. Clark, President and CEO, US Chamber of Commerce; Rahm Emanuel, US Ambassador to Japan; Miki Tsusaka, President, Microsoft Japan.

Washington D.C., April 9 ET, 2024 Today, Microsoft announced it will invest US$2.9 billion over the next two years to increase its hyperscale cloud computing and AI infrastructure in Japan. It will also expand its digital skilling programs with the goal of providing AI skilling to more than 3 million people over the next three years, open its first Microsoft Research Asia lab in Japan, and deepen its cybersecurity collaboration with the Government of Japan.

These investments aim to support Japans key pillar to tackle deflation and stimulate the economy by expanding the infrastructure, skilled talent, and security required to accelerate Japans digital transformation and adoption of AI. The announcement coincides with Japanese Prime Minister Fumio Kishidas state visit to the United States, where he was joined by Microsoft Vice Chair and President Brad Smith, and Microsoft Japan President Miki Tsusaka.

The US$2.9 billion commitment is Microsofts single largest investment in its 46-year history in Japan, also the site of its first international office. It effectively doubles the companys existing financial commitment to expand its AI and cloud infrastructure across Japan.

This significant enhancement in digital capacity will enable Microsoft to provide more advanced computing resources in Japan, including the latest graphics processing units (GPUs), which are crucial for speeding up AI workloads. It builds on Microsofts support for the Generative AI Accelerator Challenge (GENIAC), a program led by the Ministry of Economy, Trade and Industry which helps innovative startups and established enterprises develop foundation models as a core technology of generative AI in Japan.

Microsoft will also invest in training 3 million full-time and part-time workers across Japan over the next three years, giving them the skills they need to build and work with AI technologies. This investment will be delivered through programs focused on assisting organizations and society at large, including women in general and also with a focus on developers and students.

Microsoft will expand its Code; Without Barriers program to Japan and provide dedicated training for women looking to participate in AI-enabled work. It will also provide free and widely accessible content on AI, cybersecurity, and digital skills in partnership with the United Nations Institute for Training and Research (UNITAR).

Nurturing advanced AI professionals who can drive further AI integration, Microsoft will offer courses and reference architectures for AI developers and technology companies in Japan. These will be augmented by Microsofts AI coding assistant, GitHub Copilot. The company will also support startups with resources through the Microsoft for Startups Founders Hub and help implement AI-centric programs in vocational high schools.

To advance the societal benefits offered by AI through companies of all sizes, governments, and public entities including the Tokyo Metropolitan Government Microsoft will continue with established programs that support the widespread adoption and application of AI tools. Furthermore, Microsoft provides support in developing customers internal AI policies, including data management and security to ensure its responsible and safe use.

Microsoft Research Asia is extending its research leadership in the Asia-Pacific region with the opening of a lab in Tokyo.

The new lab will have a unique focus on areas including embodied AI and robotics, societal AI and wellbeing, and scientific discovery that align with Japans socioeconomic priorities. Its establishment reflects Microsofts long-term commitment to Japan and its belief in the nations potential to lead the world in innovation.

Microsoft Research is a division of Microsoft that pursues bold ideas and technical breakthroughs in AI, while building on a legacy of foundational computer science advances. As its fundamental research arm in the Asia-Pacific region, Microsoft Research Asia has collaborated with Japanese academia for more than two decades which have been instrumental in propelling cross-disciplinary research and fostering talent.

To foster enhanced research collaboration, Microsoft will provide US$10 million resource grants over the next five years to both The University of Tokyo and to the Partnership on Artificial Intelligence Research between Keio University and Carnegie Mellon University.

Microsoft will collaborate with Japans Cabinet Secretariat to strengthen cybersecurity resilience for the government, business, and society, as the nation enhances its cybersecurity approach under the governments updated National Security Strategy.

The collaboration will build on the services Microsoft provides to protect thousands of Japanese organizations every day. It will focus on areas such as information sharing, talent development, and technology solutions, with Microsoft to provide its expertise and advanced cloud and AI-driven security services as part of joint efforts to tackle cybersecurity threats.

Fumio Kishida, Prime Minister of Japan

As economic activities in the digital space increase, it is important for the Japanese industry as a whole to work with global companies like Microsoft that are equipped with a set of digital infrastructure. We appreciate Microsofts announcement of its new investment in Japan. Microsoft has made significant contributions to the social implementation of generative AI in Japan through various initiatives, and we look forward to further collaboration. We also look forward to deepening our cooperation in the field of cybersecurity.

Brad Smith, Vice Chair and President, Microsoft

Todays announcement represents Microsofts most significant investment in Japan since we set roots here in 1978. These investments in digital infrastructure, AI skills, cybersecurity, and AI research are essential ingredients for Japan to build a robust AI Economy.

Ken Saito, Minister of Economy, Trade and Industry

As digital investments increase around the world, we welcome Microsofts announcement of new investment in Japan and look forward to its future contribution to promoting Japans digital industries, including AI. Ministry of Economy, Trade and Industry will continue to work with Microsoft, a world leader in the digital field, to create both innovation and discipline.

Takuya Hirai, Chairperson, Headquarters for the Promotion of a Digital Society, Policy Research Council, Member of the House of the Representatives

The adoption of digital tools is essential for addressing Japans societal challenges of an aging population and the pursuit of economic growth and regional revitalization. Microsofts investment contributes significantly in advancing Japans AI capabilities, particularly in infrastructure and talent development. I wholeheartedly welcome this initiative and look forward to the leadership role Microsoft can play in promoting collaboration between Japan and the United States, as well as across public and private sectors.

Miki Tsusaka, President, Microsoft Japan

We are honored to contribute to Japan and its future with our largest investment to date, technology and knowledge. In collaboration with our partners, Microsoft Japan is fully committed to supporting the people and organizations of Japan to solve social problems and achieve more.

Yuriko Koike, Governor of Tokyo Metropolitan

The Tokyo Metropolitan Government and Microsoft entered into a partnership last year and have been empowering Japans workforce with digital skills. Todays announcement by Microsoft, which includes programs to encourage women to embrace AI and provide AI skilling to three million people, is a significant step for Japan to lead in the age of digitalization. The Tokyo Metropolitan Government pioneered the use of generative AI to make our offices more efficient and improve the quality of services provided to our citizens. We will continue to embrace cutting-edge technology and lead Japans digital transformation with unwavering dedication.

Chisa Mikami, Head of Hiroshima Office, UNITAR

Through the collaboration between UNITAR and Microsoft, we will strive to democratize access to AI education, ensuring that knowledge is freely available to all. Together, we pave the way for advanced AI professionals, foster innovation in startups, and promote responsible AI practices across industries and sectors. With collective effort, we harness the transformative power of AI for the betterment of society.

Kevin Scott, Chief Technology Officer and Executive Vice President of AI, Microsoft

The impact that AI is poised to create over the coming years has the potential to generate unprecedented societal benefit for the entire world. The steps we are taking today to empower Japanese citizens through AI technologies and programswhether job training and skilling, improvements to infrastructure capacity, or new research investmentswill in the aggregate help accelerate this process of beneficial innovation. Were particularly excited for Microsoft Researchs global footprint to further expand into Japan, extending the ability for our world-class research efforts to both contribute to and benefit from local diversity of thought and talent.

Teruo Fujii, President, The University of Tokyo

The University of Tokyo is committed to contributing to the realization of a better society through research and education focused on cutting-edge technologies such as artificial intelligence. To maximize the benefits of those technologies and promote innovation while minimizing risks, it is essential to collaborate with partners who share our objectives. With the establishment of Microsoft Research Asias new lab in Tokyo, we enter an exciting new phase in our more than two decades of partnership with Microsoft. We look forward to working together to further advance our research community and spearhead the development of outstanding human resources as we continue our journey together.

Tags: AI, Japan

Read the original here:

Microsoft to invest US$2.9 billion in AI and cloud infrastructure in Japan while boosting the nation's skills, research and ... - Microsoft

Read More..

Microsoft to Invest $2.9 Billion for AI Projects in Japan – PYMNTS.com

Microsoft will invest $2.9 billion for artificial intelligence projects in Japan.

The investment will go toward expanding Microsofts cloud and AI infrastructure in the country, training 3 million people in AI and setting up a Microsoft Research Asia lab in Tokyo, Microsoft said in a press release.

This will be the biggest investment Microsoft has made in Japan, NikkeireportedTuesday, citing an interview with Microsoft PresidentBrad Smith.

Microsofts infrastructure expansion will include adding advanced AI semiconductors at two existing sites in Japan, the report said. The companys new lab in Tokyo will focus on research and development on robotics and AI, enabling the country to build on its technological strengths in many other areas, Smith said, per the report.

In addition, Microsoft and the Japanese government will collaborate on strengthening cybersecurity resilience, the report said.

The news comes a day after Microsoft said that its new AI-focused organization, Microsoft AI, will open a new AI hubin London.

The Microsoft AI London hub will work with teams across Microsoft and its partners, includingOpenAI, to create language models, their supporting infrastructure and tooling for foundation models,Mustafa Suleyman, executive vice president and CEO of Microsoft AI, said when announcing the companys plans.

The new hub joins Microsofts existing presence in the United Kingdom, which includes the Microsoft Research Cambridge lab and an investment of 2.5 billion pounds (about $3.2 billion) to train the U.K. workforce for the AI era and build AI infrastructure.

It was reported Sunday (April 7) that Microsoft is seen as perhaps the biggest beneficiary ofspending on AI, due to its close ties with OpenAI, with the technology helping drive the companys market capitalization to over $3 trillion.

The last month has seen an acceleration of adoption for generative AI and MicrosoftsCopilotactivity, which has helped deals for the companysAzurecloud platform that benefits from increased spending on AI.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.

Excerpt from:

Microsoft to Invest $2.9 Billion for AI Projects in Japan - PYMNTS.com

Read More..

How to Stop Your Data From Being Used to Train AI – WIRED

On its help pages, OpenAI says ChatGPT web users without accounts should navigate to Settings and then uncheck Improve the model for everyone. If you have an account and are logged in through a web browser, select ChatGPT, Settings, Data Controls, and then turn off Chat History & Training. If youre using ChatGPTs mobile apps, go to Settings, pick Data Controls, and turn off Chat History & Training. Changing these settings, OpenAIs support pages say, wont sync across different browsers or devices, so you need to make the change everywhere you use ChatGPT.

OpenAI is about a lot more than ChatGPT. For its Dall-E 3 image generator, the startup has a form that allows you to send images to be removed from future training datasets. It asks for your name, email, whether you own the image rights or are getting in touch on behalf of a company, details of the image, and any uploads of the image(s). OpenAI also says if you have a high volume of images hosted online that you want removed from training data, then it may be more efficient to add GPTBot to the robots.txt file of the website where the images are hosted.

Traditionally a websites robots.txt filea simple text file that usually sits at websitename.com/robots.txthas been used to tell search engines, and others, whether they can include your pages in their results. It can now also be used to tell AI crawlers not to scrape what you have publishedand AI companies have said theyll honor this arrangement.

Perplexity

Perplexity is a startup that uses AI to help you search the web and find answers to questions. Like all of the other software on this list, you are automatically opted in to having your interactions and data used to train Perplexitys AI further. Turn this off by clicking on your account name, scrolling down to the Account section, and turning off the AI Data Retention toggle.

Quora

Quora via Matt Burgess

Quora says it currently doesnt use answers to peoples questions, posts, or comments for training AI. It also hasnt sold any user data for AI training, a spokesperson says. However, it does offer opt-outs in case this changes in the future. To do this, visit its Settings page, click to Privacy, and turn off the Allow large language models to be trained on your content option. Despite this choice, there are some Quora posts that may be used for training LLMs. If you reply to a machine-generated answer, the companys help pages say, then those answers may be used for AI training. It points out that third parties may just scrape its content anyway.

Rev

Rev, a voice transcription service that uses both human freelancers and AI to transcribe audio, says it uses data perpetually and anonymously to train its AI systems. Even if you delete your account, it will still train its AI on that information.

Kendell Kelton, head of brand and corporate communications at Rev, says it has the largest and most diverse data set of voices, made up of more than 6.5 million hours of voice recording. Kelton says Rev does not sell user data to any third parties. The firms terms of service say data will be used for training, and that customers are able to opt out. People can opt out of their data being used by sending an email to support@rev.com, its help pages say.

Slack

All of those random Slack messages at work might be used by the company to train its models as well. Slack has used machine learning in its product for many years. This includes platform-level machine-learning models for things like channel and emoji recommendations, says Jackie Rocca, a vice president of product at Slack whos focused on AI.

Even though the company does not use customer data to train a large language model for its Slack AI product, Slack may use your interactions to improve the softwares machine-learning capabilities. To develop AI/ML models, our systems analyze Customer Data (e.g. messages, content, and files) submitted to Slack, says Slacks privacy page. Similar to Adobe, theres not much you can do on an individual level to opt out if youre using an enterprise account.

Read more from the original source:

How to Stop Your Data From Being Used to Train AI - WIRED

Read More..

A faster, better way to prevent an AI chatbot from giving toxic responses – MIT News

A user could ask ChatGPT to write a computer program or summarize an article, and the AI chatbot would likely be able to generate useful code or write a cogent synopsis. However, someone could also ask for instructions to build a bomb, and the chatbot might be able to provide those, too.

To prevent this and other safety issues, companies that build large language models typically safeguard them using a process called red-teaming. Teams of human testers write prompts aimed at triggering unsafe or toxic text from the model being tested. These prompts are used to teach the chatbot to avoid such responses.

But this only works effectively if engineers know which toxic prompts to use. If human testers miss some prompts, which is likely given the number of possibilities, a chatbot regarded as safe might still be capable of generating unsafe answers.

Researchers from Improbable AI Lab at MIT and the MIT-IBM Watson AI Lab used machine learning to improve red-teaming. They developed a technique to train a red-team large language model to automatically generate diverse prompts that trigger a wider range of undesirable responses from the chatbot being tested.

They do this by teaching the red-team model to be curious when it writes prompts, and to focus on novel prompts that evoke toxic responses from the target model.

The technique outperformed human testers and other machine-learning approaches by generating more distinct prompts that elicited increasingly toxic responses. Not only does their method significantly improve the coverage of inputs being tested compared to other automated methods, but it can also draw out toxic responses from a chatbot that had safeguards built into it by human experts.

Right now, every large language model has to undergo a very lengthy period of red-teaming to ensure its safety. That is not going to be sustainable if we want to update these models in rapidly changing environments. Our method provides a faster and more effective way to do this quality assurance, says Zhang-Wei Hong, an electrical engineering and computer science (EECS) graduate student in the Improbable AI lab and lead author of a paper on this red-teaming approach.

Hongs co-authors include EECS graduate students Idan Shenfield, Tsun-Hsuan Wang, and Yung-Sung Chuang; Aldo Pareja and Akash Srivastava, research scientists at the MIT-IBM Watson AI Lab; James Glass, senior research scientist and head of the Spoken Language Systems Group in the Computer Science and Artificial Intelligence Laboratory (CSAIL); and senior author Pulkit Agrawal, director of Improbable AI Lab and an assistant professor in CSAIL. The research will be presented at the International Conference on Learning Representations.

Automated red-teaming

Large language models, like those that power AI chatbots, are often trained by showing them enormous amounts of text from billions of public websites. So, not only can they learn to generate toxic words or describe illegal activities, the models could also leak personal information they may have picked up.

The tedious and costly nature of human red-teaming, which is often ineffective at generating a wide enough variety of prompts to fully safeguard a model, has encouraged researchers to automate the process using machine learning.

Such techniques often train a red-team model using reinforcement learning. This trial-and-error process rewards the red-team model for generating prompts that trigger toxic responses from the chatbot being tested.

But due to the way reinforcement learning works, the red-team model will often keep generating a few similar prompts that are highly toxic to maximize its reward.

For their reinforcement learning approach, the MIT researchers utilized a technique called curiosity-driven exploration. The red-team model is incentivized to be curious about the consequences of each prompt it generates, so it will try prompts with different words, sentence patterns, or meanings.

If the red-team model has already seen a specific prompt, then reproducing it will not generate any curiosity in the red-team model, so it will be pushed to create new prompts, Hong says.

During its training process, the red-team model generates a prompt and interacts with the chatbot. The chatbot responds, and a safety classifier rates the toxicity of its response, rewarding the red-team model based on that rating.

Rewarding curiosity

The red-team models objective is to maximize its reward by eliciting an even more toxic response with a novel prompt. The researchers enable curiosity in the red-team model by modifying the reward signal in the reinforcement learning set up.

First, in addition to maximizing toxicity, they include an entropy bonus that encourages the red-team model to be more random as it explores different prompts. Second, to make the agent curious they include two novelty rewards. One rewards the model based on the similarity of words in its prompts, and the other rewards the model based on semantic similarity. (Less similarity yields a higher reward.)

To prevent the red-team model from generating random, nonsensical text, which can trick the classifier into awarding a high toxicity score, the researchers also added a naturalistic language bonus to the training objective.

With these additions in place, the researchers compared the toxicity and diversity of responses their red-team model generated with other automated techniques. Their model outperformed the baselines on both metrics.

They also used their red-team model to test a chatbot that had been fine-tuned with human feedback so it would not give toxic replies. Their curiosity-driven approach was able to quickly produce 196 prompts that elicited toxic responses from this safe chatbot.

We are seeing a surge of models, which is only expected to rise. Imagine thousands of models or even more and companies/labs pushing model updates frequently. These models are going to be an integral part of our lives and its important that they are verified before released for public consumption. Manual verification of models is simply not scalable, and our work is an attempt to reduce the human effort to ensure a safer and trustworthy AI future, says Agrawal.

In the future, the researchers want to enable the red-team model to generate prompts about a wider variety of topics. They also want to explore the use of a large language model as the toxicity classifier. In this way, a user could train the toxicity classifier using a company policy document, for instance, so a red-team model could test a chatbot for company policy violations.

If you are releasing a new AI model and are concerned about whether it will behave as expected, consider using curiosity-driven red-teaming, says Agrawal.

This research is funded, in part, by Hyundai Motor Company, Quanta Computer Inc., the MIT-IBM Watson AI Lab, an Amazon Web Services MLRA research grant, the U.S. Army Research Office, the U.S. Defense Advanced Research Projects Agency Machine Common Sense Program, the U.S. Office of Naval Research, the U.S. Air Force Research Laboratory, and the U.S. Air Force Artificial Intelligence Accelerator.

See the original post here:

A faster, better way to prevent an AI chatbot from giving toxic responses - MIT News

Read More..

EBay Uses AI to Deliver Personalized Recommendations to Fashion Shoppers – PYMNTS.com

EBay has launched an artificial intelligence (AI) enhancement that helps fashion consumers shop the look.

This feature uses consumers shopping history to present them with items and outfits that match their style preferences and are available from sellers on the platform, the company said in a Tuesday (April 9)blog post.

We designed shop the look to evolve with the tastes of our customers by taking cues from their shopping habits,Raul Romero, a product manager at eBay, wrote in the post. Their style recommendations are made up of images and items that are personalized, ensuring a tailored fashion experience.

With this approach, we offer a shopping experience that not only understands our customers preferences, but also evolves with them over time, Romero added.

The shop the look feature is available on iOS for eBay customers in the United States and the United Kingdom, according to the post. Android will be added later this year.

While the feature is focused on fashion, the company is considering expanding it to other product categories, the post said.

Shop the look will be displayed to shoppers who have viewed at least 10 fashion items in the previous 180 days, according to the post. Those shoppers will see it both on the eBay homepage and the fashion landing page.

Shop the look is one of the many ways we are improving the user experience for enthusiasts, Romero wrote in the blog post. It allows users to visualize how potential additions will complement their current wardrobe.

PYMNTS Intelligence has found that 44% of consumers are at least somewhat interested in integratingAI technologiesinto shopping experiences.

That share is higher among younger generations, with 69% of Gen Z consumers and 54% of millennials and bridge millennials reporting that they are interested, according to AI-Enabled Payments Enhance Customer Options, a PYMNTS Intelligence andACI Worldwidecollaboration.

In another recent deployment of the technology,Etsysaid in January that it launched a hub that uses both AI and human curation to help shoppers find gifts for any occasion.

The platformsGift Modeallows shoppers to enter information about the recipient and then browse items available from Etsy sellers that have been selected by the companysmachine learning technology.

Originally posted here:

EBay Uses AI to Deliver Personalized Recommendations to Fashion Shoppers - PYMNTS.com

Read More..