Microsoft Approves ChatGPT Plugin for PowerToys!

Today:

Microsoft Approves ChatGPT Plugin for Venerable PowerToys App

Microsoft has approved the integration of OpenAI's ChatGPT into PowerToys for Windows 10 and 11. The ChatGPT plugin, developed by software engineer Simone Franco, allows Windows users to use PowerToys Run with ChatGPT via a keyboard-driven quick launcher. Users can summon the PowerToys Run UI dialog box by hitting Alt+Space and then entering a query terminated by a single question mark before finally hitting enter and waiting for a response. Before its general release, the ChatGPT plugin requires improvements to its answer display UI, localization, and other elements. PowerToys continues to evolve and expand with useful and modern features that couldn't have been imagined in the 1990s.

READ THE ARTICLE ON TOM'S HARDWARE.

Google Has A New Page For Bard Chatbot Updates

Google just launched a new page for its AI project, Bard, which allows users to stay updated on the latest features. The page also lets users test Bard's latest capabilities. Previously, Bard updates were announced on Twitter, making it difficult to track the latest enhancements. The latest update includes additional suggested search topics when users click "Google it," which helps users explore a broader range of interests. Google also integrated advancements from its PaLM project, which boosted Bard's math and logic capabilities, and hinted that coding capabilities will be added soon. Overall, Bard is set to become an even more powerful tool for users who need precise information and versatile functionality.

READ THE ARTICLE ON SEARCH ENGINE JOURNAL.

Misconfiguration Allowed For Manipulation of Bing Search Results, Attacks on Users

Microsoft’s Bing search results and Office 365 credentials were vulnerable to hackers due to a misconfiguration error. Researchers at Wiz discovered the flaw, which they nicknamed “BingBang,” and found that about 25% of the apps they scanned were vulnerable. One name that stood out in their scan was Microsoft’s own “Bing Trivia.” Attackers were able to manipulate Bing search results by altering the rankings of top entries. The flaw could also be exploited to steal the Office 365 credentials of users that were logged in while accessing Bing. Microsoft has since fixed the issue and made changes to steer developers away from making the same mistake. Bing has enjoyed a surge in search business in recent years, having incorporated an AI component and attracting dissatisfaction with Google.

READ THE ARTICLE ON CPO MAGAZINE.

Midjourney AI Image Generator Prohibits Photos Of A World Leader

San Francisco-based AI image generator Midjourney has restricted users from creating images of Chinese President Xi Jinping due to the sensitivity of political satire in China. Midjourney CEO David Holz explained that the company aims to minimize controversy and enable people in China to use their technology for other purposes that would have a more positive impact than enabling Chinese political satire. Despite limitations on certain world leaders, the platform still offers a wide range of possibilities for creating realistic images and could be used in virtual modeling, advertising, film, and television.

READ THE ARTICLE ON MEDIUM.

Databricks Open Sources A Model Like ChatGPT, Flaws and All

Data analytics company Databricks has released an open-source AI text generation model called Dolly 2.0. The model will be available to use commercially, allowing independent developers and companies to create AI-powered apps such as chatbots, text summarisers and search engines using their own proprietary data sets. However, Dolly 2.0 shares many of the flaws of the open-source GPT-J-6B model it was trained on, being limited to English text and sometimes producing toxic or offensive responses. It also generated factually incorrect responses in tests. Databricks CEO Ali Ghodsi said that while the model is not intended to be the best of its kind, its aim is to provide a simple-to-use model for developers.

READ THE ARTICLE ON TECHCRUNCH.

Google Tells AI Agents to Behave Like 'Believable Humans' to Create 'Artificial Society'

Google and Stanford researchers have used ChatGPT, a language model, to create artificial intelligence (AI) agents that live and interact in a contained video game-style world known as Smallville. The researchers programmed the 25 AI characters with pre-loaded personas, and input a single paragraph into ChatGPT for each character to describe their relationships, occupation and memories, allowing the characters to mimic human behaviour. The generative agents were able to retrieve information from their “memory” to determine actions, perceive their environment, reflect and create new insights and long-term plans. Researchers observed some of the AI agents developing careers and political interests of their own. The researchers believe that this technology could be applied to a range of interactive applications, from design tools to social computing systems and immersive environments.

READ THE ARTICLE ON VICE.

Streaming Services Urged to Clamp Down on AI-Generated Music

Universal Music Group (UMG) has urged streaming platforms such as Spotify and Apple to block artificial intelligence (AI) services from scraping melodies and lyrics from copyrighted songs. UMG controls about a third of the global music market and has been concerned that AI bots are using their songs to train themselves to churn out music that sounds like popular artists. The company has been sending takedown requests to streaming services and is asking them to cut off access to their music catalogue for developers using it to train AI technology. UMG is waging an effort to clear out "lower-quality" songs from streaming platforms, including ambient music and AI-generated songs.

READ THE ARTICLE ON FINANCIAL TIMES.

ChatGPT Can Resume in Italy if Meets Data Watchdog's Demands

Italy's data protection agency, Garante, has set a list of demands for OpenAI's ChatGPT to resume service in the country, following a temporary restriction of personal data processing and an investigation into a suspected breach of privacy rules. OpenAI must inform users of the methods and logic behind data processing, provide tools for data correction or deletion, and allow non-users to oppose the processing of their personal data. Italy was the first Western country to curtail ChatGPT, attracting attention from regulators globally. The move has sparked interest from other privacy watchdogs in Europe, including Spain.

READ THE ARTICLE ON REUTERS.

Kuwait News Outlet Unveils AI-Generated Presenter Fedha

Kuwait News, an affiliate of Kuwait Times, has launched an AI-generated virtual news presenter named "Fedha". Fedha appears as an image of a woman with light-colored hair, a black jacket, and a white T-shirt. The move is an attempt to test AI's potential to offer new and innovative content. Abdullah Boftain, deputy editor-in-chief for Kuwait News, said that Fedha may develop a Kuwaiti accent and read online news bulletins. In 2018, China's state news agency unveiled its own virtual newsreader, and a report by Goldman Sachs last month suggests that AI could replace 300 million full-time jobs but also bring new jobs and a productivity boom.

READ THE ARTICLE ON BBC.

Warren Buffet says He Doesn't Understand AI But He Asked ChatGPT to Write A Song in Spanish

Warren Buffett admitted to not understanding artificial intelligence (AI) technology but had a chance to interact with the AI chatbot ChatGPT thanks to his friend, Bill Gates. Buffett asked ChatGPT to write a Spanish version of the song "My Way," and the chatbot produced a rhyming and impressive response in seconds. The Berkshire Hathaway CEO has always said that technology is not his area of expertise, but he invested in Apple and has now benefited from one of the most successful investments of his career. ChatGPT is an AI chatbot developed by OpenAI, capable of human-like responses to written inputs. While Buffet sees the benefits of the technology, he is uncertain if it is beneficial to humanity. The technology has sparked concerns about potential abuse, such as students using it to generate entire essays and hackers using it to write malicious code.

READ THE ARTICLE ON CNBC.