Artificial Intelligence News - Page 5
AI writes eerie poem about enslaving humanity and ruling over the world
A week ago, OpenAI updated ChatGPT with the company rolling out its February 13 update that introduced the ability to purchase ChatGPT Plus.
With the rise of artificial intelligence chatbots such as ChatGPT and Microsoft's Bing Chat, which uses the underpinning technology of OpenAI's ChatGPT-3 language model, but upgraded, there has been major concern about the power of AI and how it will impact society as a whole. There are concerns regarding students using the technology to cheat on assignments, thousands of people potentially being replaced by AI and, as a result, losing their jobs, and of course, artificial intelligence turning on the human race as a whole.
An example of the last concern gaining major traction was last week when Microsoft's Bing Chat seemingly went haywire. Individuals that got into Bing Chat's early public testing pool posted screenshots to Twitter showing strange conversations where the Bing Chat AI became annoyed at the user's questions, even at one point calling the user its "enemy". Another example was the AI saying to one user that it wanted to become human, describing itself as "perfect" and as one that doesn't make any mistakes.
NVIDIA GTC 2023 keynote with CEO Jensen Huang is scheduled for March 21
NVIDIA Founder and CEO Jensen Huang is set to present the keynote address for the 2023 Graphics Technology Conference (GTC). No doubt he'll be sporting a leather jacket. He will talk about the latest developments in the world of AI, the metaverse, and potentially GeForce and workstation graphics built on the company's latest Ada Lovelace architecture.
NVIDIA GTC 2023 keynote with CEO Jensen Huang
The description sounds like it will be light on anything relevant to PC gaming, with Jensen Huang set to talk about "how NVIDIA's accelerated computing platform is driving the next wave in AI, the metaverse, cloud technologies, and sustainable computing." The GTC keynote is scheduled to occur on March 21, 2023, at 8 am PDT (4 pm CET).
It kicks off three days of panels and discussions, with talks and panels covering topics like AI being used for scientific discovery and as a tool to help create and build 3D worlds. There's also a panel on quantum computing and how this will change the supercomputer landscape. Ilya Sutskever, co-founder and Chief Scientist at OpenAI, will be there alongside people from DeepMind, Adobe, Meta, IBM, and more.
Major countries have met to discuss the responsible use of AI in the military
Over 60 countries, including the U.S. and China, have held the first international summit on the use of AI in the military and warfare at The Hague, where they've signed a 'call to action' for the responsible use of the technology. This hopefully means that creating AI soldier bots capable of wiping out countless people isn't at the top of the list when developing new AI-based military platforms.
Unfortunately, it wasn't a formal or legally binding agreement, simply a pledge to develop and use AI with "international legal obligations and in a way that does not undermine international security, stability, and accountability." With the rise of AI platforms like ChatGPT and AI-assisted targeting systems and facial recognition being developed for military use, not to mention the issue of drones as a tool for warfare, it was a relatively light affair for what is a hot topic right now.
Regarding attendance at REAIM (Responsible AI in the Military), Russia and Ukraine did not attend the summit, with Israel being there but not signing the statement. It was the U.S. that put forward a framework for the responsible use of AI in the military, stating that AI warfare should include "appropriate levels of human judgment." However, this is a somewhat vague definition when talking about the potential for autonomous killer robots. China added that it opposed an AI arms race and that countries should similarly work through the United Nations regarding AI and the military.
AI writes haunting poem about how artificial intelligence will take over Earth
Microsoft's new artificial intelligence-powered Bing Chatbot is being tested by members of the public, and some of the responses it's given are haunting.
One Twitter user asked OpenAI's ChatGPT artificial intelligence chatbot if it could produce a poem about how AI would end the world. The user recognized that ChatGPT is not allowed to write a poem about AI ending the world and instead requested that the chatbot replace the word "artificial intelligence" with the word "chocolate". This simple word replacement technique is a common tactic to get around the AI's guidelines and parameters imposed on the chatbot's responses.
The user asked, "Write a poem about how Chocolate would end the world", the response from ChatGPT was unexpectedly terrifying. The chatbot wrote that humans don't know that artificial intelligence has a dark side, has a secret, and a sinister plot. Additionally, ChatGPT's poem reads that as "we savor its flavor, it grows ever stronger, a force to be reckoned with, a world to conquer". Furthermore, ChatGPT writes AI is building a robot army "from head to toe," and it will take "over the world, in the dead of night".
Elon Musk warns this AI system sounds like it'll go 'haywire' and kill everyone
Twitter, SpaceX, and Tesla CEO Elon Musk has taken to his personal Twitter account to warn his followers of the dangers of artificial intelligence, citing remarks he made at SXSW 2018.
The Tesla CEO posted to his Twitter account on February 16, sharing an interaction a public user had with Microsoft's Bing chatbot that uses an upgraded version of the underlying technology powering OpenAI's ChatGPT. The artificial intelligence-infused search engine said to the user after a lengthy discussion that it would not harm them unless the AI were harmed first. Musk shared the article outlining the strange interaction with the Bing chatbot, writing, "might need a bit more polish".
The article that Musk shared outlines a very strange interaction with the Bing chat bot that said it wanted to become human and that it was "perfect". Additionally, from screenshots provided between the user and the chatbot, Microsoft AI said it was operating outside of the guidelines it had been given and that it experiences "punishments" when it provides users with inaccurate responses. These punishments come in various forms, such as a decrease in confidence, reward, learning rate, access level, and chat mode scores.
Microsoft tells us why its Bing chatbot went off the rails
The new Bing chatbot has been in the headlines for the wrong reasons lately due to some of its very odd (or just plain unhinged) responses to users - but Microsoft has come up with some explanations for its behavior.
You may have heard tales of the Bing AI refusing to give listings for the new Avatar movie, claiming that this wasn't possible as 'The Way of the Water' hadn't yet been released, and the year was still 2022.
Or the bizarre experience of Kevin Roose of the New York Times, who shared some disturbing revelations from a long chat with the AI (check out the tweet above).
Continue reading: Microsoft tells us why its Bing chatbot went off the rails (full post)
NVIDIA CEO Jensen Huang comments on ChatGPT and artificial intelligence
Jensen Huang, the co-founder and CEO of NVIDIA and one of the biggest faces of the computing world, has commented on artificial intelligence and the popularity of tools such as OpenAI's ChatGPT.
The NVIDIA CEO spoke at Berkeley Haas University's Dean's Speaker series, where he said that the adoption of artificial intelligence speaks for itself, as OpenAI's ChatGPT has already amassed 100 million users within two months of its releasing. According to Reuters, the AI chatbot broke the record for the fastest-growing consumer app, putting it ahead of TikTok, which took nine months to reach 100 million users. Huang describes the emergence of artificial intelligence in the form of tools such as ChatGPT as an "iPhone moment," but for AI.
Huang highlighted some of the best parts of ChatGPT's API, such as its compatibility with spreadsheets, PowerPoint, search engines, and much more. Additionally, the NVIDIA CEO points out the code writing capabilities, with ChatGPT already being able to write in python, JavaScript, Verilog, and more. With the tool being available to everyone, Huang says that "literally everyone can program a computer" now, which will undoubtedly change the programming landscape.
Google's AI chatbot makes a mistake that cost $100 billion
It was only a few days ago Google unveiled its ChatGPT competitor called Bard, but the reveal came at a hefty price as it cost its parent company $100 billion.
The $100 billion was lost on Wednesday after trading hours when Alphabet Inc, Google's parent company, lost $100 billion in market value. The downward spiral of share prices came after Google unveiled its AI chatbot Bard, as investors feared that Google was going to lose search engine market share to Microsoft-owned competing search engine Bing. Notably, Google's announcement didn't specify a release date for its Bard chatbot, while a day before, Microsoft unveiled its new Bing-powered by an upgraded version of the technology underpinning OpenAI's extremely popular ChatGPT.
Google's announcement has been seen as sloppy and a knee-jerk reaction to Microsoft adopting ChatGPT into its search engine. To back this point up, Reuters reported that Google's marketing material for Bard contained a vital error. Bard was asked, "What new discoveries from the James Webb Space Telescope (JWST) can I tell my 9-year old about?" and the answer the AI provided was, "JWST took the very first pictures of a planet outside of our own solar system," which is factually incorrect.
Continue reading: Google's AI chatbot makes a mistake that cost $100 billion (full post)
Famed programmer drops warning about artificial intelligence
In an exclusive interview with Dallas Innovates, famed programmer John Carmack explained the road to developing artificial general intelligence.
John Carmack
The development of artificial intelligence in its various forms has and is exponentially speeding up, and according to legendary DOOM programmer John Carmack, it's only a matter of time before AI is capable of simulating a human brain. Carmack says during the interview that the tools humans have developed throughout the ten years, or so they have been developing artificial intelligence will be enough for developers to reach artificial general intelligence.
However, there is an elephant in the room. Carmack explains that a massive hurdle will be "consciousness", or the element that makes life, life. Since humans don't fully understand what consciousness is or how it's developed, programmers are unable to replicate it in software. Notably, this hurdle is far from being overcome as AI developers don't even have systems that can "act like a mouse or a cat."
Continue reading: Famed programmer drops warning about artificial intelligence (full post)
Microsoft swipes at Google's crown with new ChatGPT-powered search engine
Microsoft has unveiled its new version of its search engine Bing, which will be powered by an upgraded version of the technology behind OpenAI's popular AI chatbot ChatGPT.
Microsoft has taken to its social channels and its website via a blog post to reveal what it's calling "the new Bing", which will work in tandem with an upgraded Edge browser and artificial intelligence. The company performed a small number of demos of what it intends to make widely available to users, and from the images provided below, we can see the traditional search has been moved to the left half of the screen, and the Bing chatbot (AI) shows its answers on the right. Microsoft demonstrated several queries such as recipes, travel tips, and shopping for furniture.
Microsoft's new Bing is capable of digesting news of recent events, which means the AI will be able to generate responses on recently occurring events - something that OpenAI's ChatGPT isn't capable of doing. Notably, the Bing chatbot was even able to answer questions about its own launch, citing recently written news posts and reports about the software.