Higher Incidence of 'Hallucinations' in Chatbots Than Previously Thought

Meta Restricts AI-Generated Ads in Political Campaigns to Curb Misinformation

Meta has decided to restrict political campaigns and certain regulated industries from using its new generative AI advertising tools. This move aims to curb the potential misuse of AI in spreading election misinformation. The policy, part of Meta's broader advertising standards, comes amid increasing concerns about the safety and ethical use of generative AI in sensitive areas. Meta's tools, which can create varied ad content from simple text prompts, will not be accessible for ads related to housing, employment, credit, social issues, elections, politics, health, pharmaceuticals, or financial services. This decision reflects the company's cautious approach to integrating AI in areas where misinformation could have significant consequences. Meanwhile, other major tech companies like Google and Snap are also taking steps to regulate the use of AI in political contexts.

Why it matters: Meta's decision to limit the use of AI in political advertising signifies a critical step in addressing the challenges and risks associated with the evolving role of AI in shaping public opinion and political discourse.

Source

Samsung Unveils Gauss: A Privacy-Focused AI Model to Rival ChatGPT

Samsung has entered the AI arena with Gauss, a ChatGPT rival, joining the wave of companies developing AI models. Gauss, named after the renowned mathematician, features capabilities in language processing, image generation, and code creation. While its language function mirrors ChatGPT's text mode, Samsung emphasizes its application in enhancing smart device control. A key differentiator for Gauss is its on-device application, prioritizing user privacy by keeping data processing local. Currently in internal testing to boost employee productivity, Samsung plans to integrate Gauss into its products in the future.

Why it matters: Samsung's Gauss represents a significant move in the tech industry, highlighting the growing trend of integrating advanced AI functionalities directly into consumer devices while emphasizing user privacy.

Source

YouTube Tests AI-Powered Features to Enhance User Interaction and Content Discovery

YouTube is testing new generative AI features, including a comments summarizer and a conversational tool, as part of its premium package. The conversational tool, accessible through an "Ask" option on Android devices in the U.S., will allow users to ask AI-driven questions about content and receive recommendations while watching videos. The comments summarizer will organize and highlight key discussion topics in video comments, aiding viewers and creators in navigating large comment sections. These features, currently in English and available on an opt-in basis, are part of YouTube's broader experimentation with AI across its platform, including AI-powered ads, music, and video creation tools. YouTube is starting these tests on a small scale, seeking user feedback to refine the experience.

Why it matters: YouTube's integration of AI features underscores a significant shift towards enhancing user engagement and content discovery on digital platforms, potentially reshaping how viewers interact with video content and creators gather audience insights.

Source

Higher Incidence of 'Hallucinations' in Chatbots Than Previously Thought

A study by Vectara, a start-up founded by ex-Google employees, reveals that chatbots, including those from major AI companies, often 'hallucinate,' or generate incorrect information. This happens at varying rates, from 3% for OpenAI's technology to a high of 27% for Google's Palm chat. The study, which focused on chatbot summaries of news articles, found consistent invention of information. This issue isn't confined to chatbots; it's a broader AI problem affecting systems that generate text or images. Given the rising use of chatbots in critical sectors like healthcare and finance, this tendency to hallucinate poses risks of misinformation and incorrect advice. The study calls for greater transparency from AI companies about their technology's limitations and urges users to be vigilant and fact-check AI-generated information.

Why it matters: This study sheds light on a significant flaw in AI chatbots, raising concerns about their reliability and the potential risks of misinformation in critical decision-making areas, thereby emphasizing the need for improved accuracy and responsible use of AI technology.

Source

Xbox Partners with Inworld AI to Revolutionize Game Development with Generative AI Tools

Xbox is teaming up with Inworld AI to revolutionize the gaming industry through Generative AI. This multi-year partnership aims to enhance game development by integrating advanced AI tools into narrative and dialogue creation. Leveraging Inworld AI's expertise in character development and Microsoft's robust cloud-based AI solutions, including Azure OpenAI Service, this collaboration seeks to provide game creators with an AI design copilot and an AI character runtime engine. These tools will enable developers to generate detailed scripts, dialogue trees, quests, and dynamically-generated stories, thus enriching player experiences. Adhering to principles of meaningful innovation, empowerment, and responsibility, Xbox and Inworld AI are committed to creating responsible AI tools accessible to developers worldwide. This partnership represents a significant leap in game development, aiming to simplify the creative process and open up new possibilities for engaging and dynamic gaming experiences.

Why it matters: The partnership between Xbox and Inworld AI signifies a transformative step in game development, offering powerful AI tools that could redefine how stories and characters are created, potentially leading to more immersive and innovative gaming experiences.

Source

Google's AI-Powered Search Experience Goes Global, Enhancing Interactivity in Over 120 Countries

Google's AI-powered search experience, Search Generative Experience (SGE), is expanding globally to over 120 countries and territories, now supporting additional languages including Spanish, Portuguese, Korean, and Indonesian. This marks a significant step in Google's AI journey, following its initial launch in the U.S., India, and Japan. SGE, similar to Microsoft's Bing Chat, allows users to interact with an AI in natural language, providing well-formed answers with cited references instead of just links. The global rollout includes new features like the ability for users to ask follow-up questions directly on the search results page, improved translation options with selectable word meanings, and expanded educational content with definitions in various subject areas. This expansion not only enhances user experience but also ensures the relevance of Google's ads business in the evolving AI-powered search landscape.

Why it matters: Google's SGE global expansion represents a major advancement in making AI-powered search more accessible worldwide, offering users a more intuitive and informative way to interact with information on the web.

Source

Mozilla Launches Fakespot Chat: An AI Tool to Detect Fake Reviews and Enhance Online Shopping

Mozilla has unveiled Fakespot Chat, its inaugural AI model and chatbot, designed to counter the growing issue of fake online reviews. Integrated as a browser extension, Fakespot Chat specifically targets Amazon product reviews, helping users discern their authenticity. Beyond identifying fake reviews, the chatbot assists users in product research and evaluation, offering interactive guidance and even suggesting questions to inquire about specific products. This development represents Mozilla's proactive approach in enhancing online shopping experiences and addressing the increasing challenge of deceptive online practices.

Why it matters: Fakespot Chat marks a significant step in using AI to improve the trustworthiness and reliability of online shopping, offering users a tool to navigate the often murky waters of product reviews and make more informed purchasing decisions.

Source

Humane Ai Pin: Ex-Apple Employees Launch a Groundbreaking Wearable AI Device

The Humane Ai Pin, designed by former Apple employees, is a groundbreaking wearable AI device that projects information onto your hand, potentially revolutionizing how we interact with technology. This screenless device, integrating OpenAI’s models, offers a unique user experience by projecting text and buttons onto the user's hand, and is controlled through hand gestures and voice commands. It functions independently of a phone, hinting at a future where such devices might replace smartphones. The Ai Pin's capabilities include real-time voice translation, music playback, call and text handling, and question answering. Priced at $699, it comes with battery boosters, a charging pad and case, and an option for cellular coverage at an additional cost. The blend of Humane’s Cosmos OS with OpenAI’s technology marks a significant shift in the wearable tech landscape.

Why it matters: The Humane Ai Pin represents a leap in wearable technology, potentially setting a new standard for how we interact with digital information and signaling a shift towards more intuitive, hands-free, and screenless devices.

Source

Something To Try - Build Your Own GPT Apps🛠️

If you’re like me, you’re a busy a business professional, who does not know how to code, BUT you want to try out some of the newest technology (and maybe get more out of ChatGPT). Well, ChatGPT just made it much simpler to do so! If you have ChatGPT Plus subscription, you can now build your own customized GPT that can help you get laser-focus on specific outputs. We are going from a Generalist to more of a Specialist approach!

How to get started

Access ChatGPT and select 'Explore' located at the top left of your screen. This action will navigate you to a page displaying all your existing GPT apps.

[Optional but Recommended] Test one of the GPTs made by OpenAI

To grasp how GPTs function, choose an app from the Explore screen's list and experiment with it.

Build your own GPT

1. Navigate back to the Explore screen and select 'Create a GPT' located at the top.

2. In the new window, use the left side of the screen to input a prompt. Tell ChatGPT about the type of GPT you want to create and its intended purpose.

3. ChatGPT will pose some follow-up questions to further fine-tune your GPT.

4. After completion, your GPT will be accessible on the right side of the screen.

For example, in the image above, I instructed ChatGPT to develop a 'Product Design Expert' GPT. This GPT allows users to upload screenshots of their apps or websites and receive design improvement suggestions. I designed it to be user-friendly and non-technical.

Remember, you have the option to save and share your GPTs with others.