Category Archives: Artificial Intelligence

Artificial Intelligence

Who Is Mira Murati? The Story Behind Her Brief CEO Role At OpenAI

During her time at Tesla and Leap Motion, Murati was introduced to AI. In an interview with Wired, she expressed her strong belief in the potential of AGI, stating that she wanted to be at the forefront of this groundbreaking technology after experiencing its application in the real world at her previous positions. At the time, OpenAI and DeepMind were the primary organizations exploring the technology, and she aspired to be involved from an early stage. In a conversation with Fortune, she explained that she chose OpenAI because she was aligned with the company’s mission to ensure that artificial general intelligence benefits all of humanity.

Upon joining OpenAI, Murati found a nonprofit research lab with only a small team of employees, which is a stark contrast to its current state with over 700 employees and a few investors supporting the project. She believed that transitioning to a for-profit model was necessary, as large-scale research requires substantial funding.

She noted that the company’s culture has evolved significantly since her arrival, with a greater focus on product development than pure research. However, Murati emphasized to Wired the importance of preserving the nonprofit’s mission.

Murati also advocates for public testing as a means to enhance generative AI. In an interview with FastCompany, she highlighted the significance of ensuring that the development of this technology is progressing in the right direction, rather than in isolation.

Artificial Intelligence, News, Technology

Microsoft Could Bring Its Copilot AI To Windows 10 After All

While Microsoft plans to end support for Windows 10 in October 2025, the company is considering adding Copilot to the older operating system. Despite the impending end of support, Windows 10 is still used by over a billion devices worldwide, compared to the 400 million active devices using Windows 11. By integrating Copilot with Windows 10, Microsoft aims to reach a much larger audience.

This move not only targets Microsoft’s direct customers but also encourages third-party developers to create plugins and applications for Copilot. By increasing the user base for Copilot, Microsoft hopes to attract more developers to code for its AI instead of rival LLMs like ChatGPT and Google Bard, as explained by Bowden. The wider reach of Windows 10 is seen as a significant incentive for developers to prioritize working on Copilot compatibility.

Despite the plan to end support for Windows 10 in the next two years, Microsoft is currently discussing the possibility of extending its support. In the meantime, Bowden’s report suggests that Windows 10 may receive new features and services that are enjoyed by Windows 11 users, including Copilot. According to Bowden, Copilot is just one of several features that the company intends to bring to the older operating system.

Artificial Intelligence, News, Technology

These New AI Scams Are On The Rise, Here’s How To Spot Them Before It’s Too Late

Instances of AI tools being used for fraudulent activities are increasing. In a recent case, a Chinese man fell victim to a cybercriminal who used an AI face-swapping tool to impersonate the victim’s friend. The criminal persuaded the man to transfer over half a million dollars. This incident was reported by Reuters.

CBS News also covered the growing trend of audio AI tools that can replicate someone’s voice. Scammers are using this technology to pose as individuals in distress. CNN reported on a similar incident where scammers attempted to deceive a mother by using a fake voice of her daughter, claiming that she had been kidnapped. Another case, reported by Insider, involved a father receiving a phone call informing him about a serious accident involving his son.

Criminals often exploit these situations by sending fake media to loved ones and demanding money for emergency assistance. This is a modern version of imposter scams, which have been around for a while. According to a report by the FTC in February 2023, American citizens lost approximately $2.6 billion to this type of fraud in 2022. However, the emergence of generative AI has significantly raised the stakes.

Artificial Intelligence, Electric Vehicles, Energy, News, Robotics, Technology

White House Announces 31 U.S. Tech Hubs to Drive Innovation

The White House has officially designated 31 technology hubs in the United States to foster innovation and support the growth of various industries. These hubs cover different categories, including autonomous systems technology and biotechnology applied in healthcare and pharmaceuticals. The tech hubs are located across the country, spanning states such as Montana, Rhode Island, Oklahoma, Massachusetts, Colorado, Illinois, Ohio, Wisconsin, Indiana, Virginia, New Hampshire, Missouri, Kansas, Maryland, Alabama, Pennsylvania, Delaware, New Jersey, Louisiana, Idaho, Georgia, South Carolina, New York, Florida, Oregon, Texas, Maine, Washington, Vermont, and even Puerto Rico.

Several federal agencies, including the USDA and the Transportation Department, will provide assistance to these tech hubs when appropriate. Gina Raimondo, the U.S. Secretary of Commerce, expressed the importance of these consortia, stating, “Each of these consortia will help us ensure the industries of the future — and their high-paying jobs — start, grow, and remain in the United States.” It is worth noting that the selection process for these tech hubs involved entities applying for the designation, with the 31 chosen from a pool of approximately 400 applicants. The grants allocated to support these tech hubs will amount to nearly $500 million.

Artificial Intelligence, News

DALL-E 3 Is Now Built Right Into ChatGPT Plus: Here’s How To Use It

DALL.E-3 image generator by OpenAI Nadeem Sarwar / SlashGear

OpenAI’s latest text-to-image generation model, DALL.E-3, is now available to all paid ChatGPT users. The updated model, which can now understand more nuanced text prompts, was announced last month. Enhanced image generation capabilities are definitely a bonus, as is the ability to create visuals based on detailed instructions. However, one of the biggest changes was that OpenAI baked DALL.E-3 right into ChatGPT, which means you can summon the image generator AI right within the ChatGPT app.

While that is impressive in itself, it’s the sheer convenience here that really stands out. You no longer have to jump between two apps to get the job done. Moreover, in the ChatGPT mobile app or web interface, you can simply ask the AI model to write a better, more detailed prompt to generate an image based on your circumstantial requirements, which actually saves a bucketload of creative brainstorming efforts.

Beta-testing for DALL.E-3 began on October 16, and now, it is widely available for all ChatGPT Plus and ChatGPT Enterprise subscribers. Interestingly, ChatGPT isn’t the first tool of its kind to get DALL.E-3 superpowers. In the first week of October, Microsoft integrated OpenAI’s image latest image generator tool within Bing Chat, letting users create images while interacting with the AI chatbot.

It’s all about the ease

If all that’s mentioned above sounds convenient, here are the steps you need to follow:

  1. As mentioned above, you need a ChatGPT Plus or Enterprise subscription to access DALL.E-3 within the ChatGPT app. ChatGPT Plus costs $20 per month, while the pricing of the Enterprise tier varies based on the organizational requirements.
  2. Once the subscription is active, users can simply ask the AI to help with image generation tasks in the ChatGPT app. A core advantage of using DALL.E-3 within ChatGPT is that the prompts rely more on situational awareness and guesswork instead of depending solely on the text prompts given by the user.
  3. For example, you can simply tell ChatGPT that you need a logo inspired by a lion for a safari adventure package, and the integrated image generator will do the rest. You don’t need to go into the details of the visual representation the way you would while using the standalone DALL.E-3 engine or any other AI-drive image generator tool of its ilk.

Alongside Dall.E-3’s upgraded image generation capabilities, OpenAI has also enhanced the safety guardrails so that it doesn’t generate offensive or graphically explicit imagery. Moreover, it will no longer ape the style of living artists, a move that supposedly tries to shield OpenAI from copyright claim heat.

Artificial Intelligence, Technology

Why Windows Copilot Has Users A Little Underwhelmed

A user on the Windows 11 subreddit, meanwhile, asked their fellow posters to keep a running list of Copilot’s useful features so far. The top comment listed off a bunch of attempted prompts, with only three having been successful: Opening apps, toggling light and dark modes, and starting a focus session. Everything else, such as closing apps, searching within documents or emails, installing apps, and personalizing desktop or taskbar settings, was off-limits. The top reply to that comment clarified that Copilot could only open apps, not any other files, even files generated by Copilot.

Another commenter on the same post believed that he had gotten Copilot to “output all of its internal functions.” He gave the caveat that there could be more, but all of the ones he listed worked. That list included searching the web, creating AI-generated art, adjusting audio volume, launching apps, screencasting, launching a troubleshooter, opening the device manager, muting the audio device, turning on Bluetooth, changing the desktop color theme, turning on “do not disturb” mode, setting volume and focus sessions, changing desktop wallpaper, and snapping windows.

All told, that’s not a lot. It seems like it will be a while before Copilot does everything in Windows that users are expecting of it.

Artificial Intelligence, News

Google Introduces AI Image Generation through Search Bar

Google has announced a new feature that allows users to generate AI images directly from the search bar. As part of this update, Google is also providing a new tool for writing inspiration. An example given by Google is a scenario where someone is researching contractors to convert their garage into an office. In this case, the person can simply type, “Write a note to a contractor asking for a quote to turn my garage into a home office,” and the AI will generate a draft to inspire them.

Once the draft is generated, users have the option to make some edits. They can choose to shorten, lengthen, or make the draft more casual. Once satisfied with the AI’s work, users can export the draft to Google Docs or Gmail and make further edits themselves.

However, it’s important to note that this feature, known as the SGE program, is currently only available to a limited number of users in the U.S., India, and Japan. Additionally, access to the image generation capabilities is restricted to U.S. Google account holders who are 18 years or older.

Source: SlashGear

Artificial Intelligence, Internet, News, Technology

ChatGPT Introduces Web Browsing, But Only for Paid Subscribers

It’s worth noting that not everyone can use ChatGPT’s web browsing feature for free. Currently, OpenAI has limited access to the feature to ChatGPT Plus and Enterprise users who pay a monthly fee for an enhanced version of the chatbot. Whether it’s worth the price depends on your specific needs. OpenAI highlights several use cases for this updated chatbot version.

One notable improvement is the ability to provide users with real-time information on a wide range of topics, including news, stock market updates, sports scores, and weather forecasts. Unlike traditional search engines, ChatGPT doesn’t just present data; it can also draw conclusions and offer suggestions for the best course of action.

This capability can be extremely valuable for professionals working in finance, research, and data analysis, where accurate and current information is crucial. It can also assist content creators in producing engaging content that incorporates the latest trends and developments. While it’s early to judge its effectiveness in this new form, ChatGPT is likely to offer users more informative, helpful, and up-to-date responses.

Artificial Intelligence, News

Meta’s Latest Llama AI Tool Can Help You Correct Code Errors

Meta has trained a tool called Code Llama on code snippets to assist developers in correcting mistakes. According to the company, Code Llama outperformed its competitors in coding benchmarks like HumanEval. Meta is launching Code Llama in three sizes: 7 billion parameters, 13 billion parameters, and 34 billion parameters. The smaller model is suitable for less demanding tasks, while the larger models require more hardware but offer more capabilities.

For example, the basic 7B version can run on a machine with a single GPU and is ideal for tasks such as code completion with low latency. The 13B model has a more powerful fill-in-the-middle (FIM) feature, and the 34B variant provides advanced code assistance, including code generation, block insertion, and debugging, for expert users who have the necessary hardware.

Meta Llama code AIMeta

Additionally, Meta has developed two other versions of Code Llama specifically designed for different coding environments. Code Llama Python focuses on assisting with Python, which is widely used for AI and machine learning tasks. The other version, Code Llama — Instruct, is tailored for natural language prompts and caters to non-experts who want to generate code. While it’s not perfect, Meta claims that Code Llama provides safer responses compared to its competitors. Code Llama is available for download on Meta’s website and can also be found on GitHub. Meta has also published a research paper that delves deeper into the details of this AI tool for code generation.