IBL News

Author: IBL News

  • Adobe Unveils Firefly, a Family of Creative Generative AI Models Coming to Its Products

    Adobe Unveils Firefly, a Family of Creative Generative AI Models Coming to Its Products

    IBL News | New York

    Adobe unveiled last week a host of new, creative generative AI called Firefly, which focused on the creation of images and text effects.

    The first applications that will include Adobe Firefly integration  — now in beta — will be Adobe Express, Adobe Experience Manager, Adobe Photoshop, and Adobe Illustrator.

    The company will also introduce a “Do Not Train” tag for creators who do not want their content used in model training; the tag will remain associated with content wherever it is used, published, or stored.

    ”With Firefly, Adobe will bring generative AI-powered ‘creative ingredients’ directly into customers’ workflows, increasing productivity and creative expression for all creators from high-end creative professionals to the long tail of the creator economy,”’ said David Wadhwani, President of Digital Media Business of Adobe.

    Adobe said that it is planning to make Firefly available via APIs on various platforms to enable customers to integrate into custom workflows and automation.

    Update May 29, 2023:

    • New Photoshop

  • Quora Released its Poe Chatbot, A Tool That Includes GPT-4, Claude and ChatGPT

    Quora Released its Poe Chatbot, A Tool That Includes GPT-4, Claude and ChatGPT

    IBL News | New York

    Quora released its Poe chatbot, a tool powered by models from OpenAI and Anthropic.

    In a blog post announcement, its CEO Adam D’Angelo explained that the company was considering more language models for the future: “Different models will be optimized for different tasks, they will represent different points of view, or they will have access to different knowledge.”

    Poe is currently the only consumer internet product with either Claude or Claude+ available right now. It is offered now for free (download it here) as a free app with limited features. The paid tier costs $9.99 per month and will give you access to both GPT-4 and Claude+.

    Meanwhile, OpenAI is charging $20 for ChatGPT Plus.

    Poe, which stands for “Platform for Open Exploration,” is Quora’s attempt to democratize access to AI chatbots and foster curiosity and learning among users.

    Poe allows users to ask questions and have conversations with various AI-powered chatbots, including GPT-4, Claude+, Claude, ChatGPT and more. This way, users can try different chatbots and easily switch between different bots, comparing their responses.

    Users can also request new personalities to be added to Poe’s roster of bots.

  • Zoom Adds to Its IQ Smart Companion New Features Provided by OpenAI

    Zoom Adds to Its IQ Smart Companion New Features Provided by OpenAI

    IBL News | New York

    Zoom is adding new AI-powered features to its video conferencing app to be able to compete with Microsoft Team, Google Workspace, and Salesforce’s Slack.

    In a blog post published on Monday, the company announced a partnership with OpenAI that will add more tools to its proprietary Zoom IQ AI-powered assistant.

    The new Zoom IQ can summarize what users have missed in real time and ask further questions. If they need to create a whiteboard session for their meeting, Zoom IQ can generate it based on text prompts.

    Once the session ends, Zoom IQ summarizes the meeting and posts that recap to Zoom Team Chat, even suggesting actions for owners to take on.

    Zoom IQ chat also drafts and rephrases responses and sends follow-ups with customers over email.

    In addition, the company also launched Zoom IQ for Sales, which uses conversational intelligence to improve seller performance.

  • Databricks Launches Dolly, an Open Sourced LLM Clone of Stanford’s Alpaca Model

    Databricks Launches Dolly, an Open Sourced LLM Clone of Stanford’s Alpaca Model

    IBL News | New York

    The big data analytics firm Databricks open-sourced last week a new AI model called Dolly, along with all of its training code and instructions on how to recreate it.

    “Dolly is a cheap-to-build LLM (large language model) that exhibits a surprising degree of the instruction following capabilities exhibited by ChatGPT,” the company announced in a blog post.

    The model underlying Dolly has only 6 billion parameters, compared to 175 billion in GPT-3. It is only two years old, “making it particularly surprising that it works so well.”

    In February 2023, Meta released the weights for a set of high-quality language models called LLaMA for academic researchers.

    In March 2023, Stanford University built the Alpaca model, which was based on LLaMA, but tuned on a small dataset of 50,000 human-like questions and answers.

    Databricks evaluated Dolly on the instruction-following capabilities described in the InstructGPT paper on which ChatGPT is based.

    Dolly — named after Dolly the sheep, the first cloned mammal — is an open-source clone of an Alpaca, inspired by a LLaMA.

    Instead of creating its own model from scratch or using LLaMA, Databricks took a much older and open-source LLM called GPT-J, which was created by EleutherAI several years earlier.

    GTP-J was the foundation on which Dolly was built.

    Databricks was able to take the EleutherAI model and make it “highly approachable” simply by training it with a small, 50,000-word dataset in less than three hours using a single machine.

    “This shows that the magic of instruction following does not lie in training models on gigantic datasets using massive hardware,” Databricks explained.

    “Rather, the magic lies in showing these powerful open-source models specific examples of how to talk to humans, something anybody can do for a hundred dollars using this small 50.000 dataset of Q&A examples.”

     

    “It exhibits many of the same qualitative capabilities, including text generation, brainstorming, and open Q&A.”

    “We believe models like Dolly will help democratize LLMs, transforming them from something very few companies can afford into a commodity every company can own and customize to improve their products,” Databricks said.

  • Microsoft Search Engine Bing Adds DALL-E’s Image AI Creator

    Microsoft Search Engine Bing Adds DALL-E’s Image AI Creator

    IBL News | New York

    Microsoft announced this week that its new AI-enabled Bing chat will allow users to generate images — one of the most searched categories, second only to general web searches.

    Powered by an advanced version of OpenAI’s DALL-E model, this new feature called Bing Image Creator, now offered to a few users, will allow users of Microsoft’s Edge browser to create an image by typing a description, providing contexts like location or activity, and choosing an art style.

    It means that Bing now can generate both written and visual content within a chat.

    “It’s like your creative copilot. Just type something like “draw an image” or “create an image” as a prompt in chat to get creating a visual for a newsletter to friends or as inspiration for redecorating your living room,” said Yusuf Mehdi, Corporate Vice President & Consumer Chief Marketing Officer of Microsoft.

    In addition to Bing Image Creator, Microsoft is rolling out new AI-powered Visual Stories — similar to Instagram stories — and updated Knowledge Cards, AI-generated infographics will be the new Bing and Edge preview, as shown below.

    The preview experience of Image Creator is now available at bing.com/create for Bing users in English.

    knowledge card showing information about corgis

    knowledge card showing information about Rio in Brazil

     

     

  • Google Provides Limited Access to Bard, Its “Early Experiment Chatbot”

    Google Provides Limited Access to Bard, Its “Early Experiment Chatbot”

    IBL News | New York

    Yesterday, Google began providing limited access to Bard, its rival to ChatGPT, to selected users in the United States and the United Kingdom. The date for full public access has not been announced yet. Users can join a waitlist for Bard at bard.google.com.

    Google stresses that Bard is a “complement to search”, not a replacement, given the tendency of these bots to invent information or hallucination, as users notice in OpenAI’s ChatGPT and Microsoft’s Bing chatbot. It all underscores the experimental nature of the technology.

    The search giant describes Bard as “an early experiment, intended to help people boost their productivity, accelerate their ideas, and fuel their curiosity.”

    In a demo for The Verge, Bard — based on Google’s AI language model LaMDA — generated three responses to each user query, with minimal variation in their content. Underneath each reply is a prominent “Google It” button.

    Bard’s interface is festooned with disclaimers such as “Bard may display inaccurate or offensive information that doesn’t represent Google’s views.”

    Bard lacks Bing’s clearly labeled footnotes, which Google says only appear when it directly quotes a source like a news article.
    A GIF showing Bard responding to a query about how to introduce your daughter to flyfishing.

    According to a report in TechCrunch, Google’s Bard is lagging behind OpenAI’s GPT-4, and Anthropic’s Claude in a head-to-head comparison on a few example prompts.

    “Overall GPT-4 is unambiguously ahead of the others, though depending on the context Claude and Bard can be competitive. Importantly, however, both Claude and Bard gave factually incorrect answers at times, and Bard even made up a citation to support its assertion about GDPR enforcement.”

  • OpenAI Starts to Roll Out Plugins in ChatGPT

    OpenAI Starts to Roll Out Plugins in ChatGPT

    IBL News | New York

    OpenAI started to roll out plugins in ChatGPT with a small set of users. These plugins, now in alpha mode, extend the bot’s functionality by connecting ChatGPT to third-party applications.

    The plugins enable ChatGPT to interact with APIs defined by developers, performing a wide range of actions, such as:

    • Retrieve real-time information; e.g., sports scores, stock prices, the latest news
    • Retrieve knowledge-base information; e.g., company docs, personal notes
    • Perform actions on behalf of the user; e.g., booking a flight, ordering food

    They are available to developers on the waitlist.

    The stated goal of OpenAI is to build a community of plugin developers.

    To foster the creation of new plugins, OpenAI has open-sourced a “retrieval” plugin that enables ChatGPT to access snippets of documents from data sources like files, notes, emails, or public documentation by asking questions in natural language. [Documentation] 

    The first plugins have been created by Expedia, FiscalNote, Instacart, KAYAK, Klarna, Milo, OpenTable, Shopify, Slack, Speak, Wolfram, and Zapier.

    The OpenTable plugin allows the chatbot to search across restaurants for available bookings, for example, while the Instacart plugin lets ChatGPT place orders from local stores. Zapier connects with apps like Google Sheets, Trello, and Gmail to trigger productivity tasks.

    The most intriguing plugin would be OpenAI’s first-party web-browsing plugin, which would allow ChatGPT to draw data from around the web, as its knowledge is limited today to info prior to 2021.

    An AI startup called WebGPT built a plugin with access to the live web.

    Beyond the web plugins, OpenAI released a code interpreter for ChatGPT that provides the chatbot with a working Python interpreter in a sandboxed, firewalled environment along with disk space.

    It supports uploading files to ChatGPT and downloading the results; OpenAI says it’s particularly useful for solving mathematical problems, doing data analysis and visualization, and converting files between formats.

  • Canva Unveils New AI-Powered Design Tools

    Canva Unveils New AI-Powered Design Tools

    IBL News | New York

    Visual communication platform Canva, which touts 120 million users, announced today a host of new AI-powered design tools and a new Brand Hub.

    “Ten years ago, we launched Canva driven by a vision to empower everyone to design anything, no matter their skills or experience,” said Canva co-founder and CEO Melanie Perkins in a statement. “Today, the huge advancements we’re seeing with artificial intelligence have given us the opportunity to once again reimagine the design process.”

    • Assistant lets users search for design elements and gives recommendations on graphics and styles that match existing designs.

    It also provides access to Magic Write, the platform AI copyright assistant launched in December used for writing website copy, presentation summaries, and ideas.

    • Canva has also launched a new way to automatically generate presentations.

    • A new “Magic Edit” feature lets add or replace anything in an image. A “Magic Eraser” feature allows users to clean up unwanted details in images by brushing over the area.

    Canva Magic Presentations

    • Canva’s new “Beat Sync” automatically matches video footage to a soundtrack, along with a new “Translate” feature that automatically translates text in designs to over 100 different languages.

    In addition to the new AI-powered tools, Canva is introducing a set of new features that it says users have requested:

    — “Draw” lets users sketch a simple shape, like a circle, and then have it transform into a polished circle.

    — “Layouts” allows users to add content to a page and get recommendations for layout ideas for a design

    — “Styles” gives users the ability to browse suggested color palettes and fonts.

    — “Layers” tool can be used to see all elements in one place, including text, shapes, images, and video.

    Canva Layouts

    Canva brand kit

    https://www.youtube.com/watch?v=FSJTA9BG-ds

  • GitHub Launches an Upgraded AI Assistant for Developers, Copilot X, With GPT-4

    GitHub Launches an Upgraded AI Assistant for Developers, Copilot X, With GPT-4

    IBL News | New York

    The GitHub Copilot programming app, owned by Microsoft, is being upgraded to use OpenAI’s latest GPT-4 language model, as the owner company Microsoft announced yesterday.

    In addition, GitHub is adding a chat and voice feature, so developers can give instructions or ask questions using their voice. It also brings Copilot to pull requests, the command line, and docs to answer questions on projects.

    The new version, called Copilot X, can provide explanations of what segments of code are meant to do, create ways to test the code and propose fixes for bugs.

    Developers can sign up for a waitlist to preview the new service.

    “With AI available at every step, we can fundamentally redefine developer productivity; we are reducing boilerplate and manual tasks and making complex work easier across the developer lifecycle,” wrote Thomas Dohmke from GitHub. “GitHub Copilot is already writing 46% of code and helps developers code up to 55% faster.”

    A graphic showing how many developers and companies have already used GitHub Copilot and how it's helping improve productivity and happiness.

     

  • Anthropic Launched ‘Claude’ A More Ethical Bot That Competes With ChatGPT

    Anthropic Launched ‘Claude’ A More Ethical Bot That Competes With ChatGPT

    IBL News | New York

    Anthropic, a Google-backed startup co-founded by ex-OpenAI employees, launched a chatbot called Claude that competes directly with offerings from Microsoft Corp-backed OpenAI, the creator of ChatGPT.

    The company, run by a former OpenAI researcher, Dario Amodei, said that it wants to be more ethical than ChatGPT, producing less harmful outputs.

    Anthropic explained it gave Claude a set of principles at the time the model is “trained” with vast amounts of text data. Rather than trying to avoid potentially dangerous topics, Claude is designed to explain its objections based on its principles.

    The pricing of Claude has not been detailed yet, although two versions are available via an API, Claude and a faster Claude Instant.

    Like ChatGPT, Claude can be instructed to perform a range of tasks, such as answering questions, writing, summarizing, translating, explaining complex concepts, searching across documents, editing legal contracts, and coding.

    Anthropic has been quietly testing Claude with launch partners, including Robin AI, AssemblyAI, Notion AI, Quora, and DuckDuckGo.

    Both Quora’s Poe chatbot and DuckDockGo use a combination of GPT-4 and Claude+, currently the two most advanced language models in the market.

    Experts said that Claude is reportedly worse at math and a poorer programmer than ChatGPT, and it equally hallucinates.

    Recently, Google pledged $300 million in Anthropic for a 10% stake in the startup. Under the terms of the deal, Anthropic agreed to make Google Cloud its “preferred cloud provider.”

    Large language models are algorithms that are taught to generate text by feeding them human-written training text. In recent years, researchers have obtained much more human-like results with such models by drastically increasing the amount of data fed to them and the amount of computing power used to train them.