
Sign up to save your podcasts
Or


John Swicki provides a news roundup of recent developments in AI. He starts by discussing Canva's new feature called Magic Studio, which includes text-to-design, text-to-video, and text-to-presentation capabilities. He is impressed by the range of features and the ease of use offered by Canva. Next, he talks about Arcmax, a web browser that has integrated AI components. He highlights features such as a five-second preview of web pages, smartly renamed downloads, and the ability to ask questions based on the content of a web page. John also mentions Google's addition of Bard to Google Assistant, which will allow users to interact with Assistant through text, voice, and photos. He believes this integration has the potential to make Assistant even more powerful. He briefly mentions Snipped, a podcast app that now offers podcast summaries, and Bing Chat, which now allows users to use Dalle3 for image generation.
Key Takeaways:
* Canva's Magic Studio offers a range of AI-powered features, including text-to-design, text-to-video, and text-to-presentation capabilities.
* Arc Max, a web browser, has integrated AI components such as a five-second preview of web pages and smartly renamed downloads.
* Google is adding Bard to Google Assistant, allowing users to interact with Assistant through text, voice, and photos.
* Snipped, a podcast app, now offers podcast summaries, making it easier for users to get a quick overview of lengthy episodes.
* Bing Chat now allows users to use Dalle3 for image generation.
Links:
https://www.canva.com/newsroom/news/magic-studio/
https://blog.google/products/assistant/google-assistant-bard-generative-ai/
https://www.anthropic.com/index/evaluating-ai-systems
https://www.snipd.com/promo/summary
https://blogs.bing.com/search/october-2023/DALL-E-3-now-available-in-Bing-Chat-and-Bing-com-create-for-free
https://arc.net/e/D25B2EEA-7506-4850-A169-3B2A00802889
Hosted on Acast. See acast.com/privacy for more information.
We discus the recent announcements made by OpenAI regarding the updates to their Chat GPT and the introduction of DALL·E 3, their text image system. He highlights the improvements in image generation and the integration of Dolly Three into ChatGPT. He also explores the new features of back-and-forth conversations and voice capabilities in ChatGPT, emphasizing the impressive quality of the voices. John expresses his excitement about the possibilities these updates bring and the potential for more robust conversations and problem-solving. He concludes by mentioning OpenAI's upcoming developer conference and the anticipation surrounding these updates.
Key Takeaways:
* OpenAI has announced significant updates to ChatGPT and the introduction of Dolly Three, a text image system.
* DALL·E 3 improves image generation and brings it up to par with other platforms.
* ChatGPT now offers back-and-forth conversations and voice capabilities with impressive quality.
* The new features allow for more interactive and problem-solving conversations.
* OpenAI's updates have generated excitement and anticipation among users.
Quotes:
* "DALL·E 3 is definitely right up there with some of the best image generation platforms I've seen so far."
* "The quality of the voices in ChatGPT is really quite good. I couldn't tell the difference between the AI voice and the human voice in the demo."
* "The back-and-forth conversations and problem-solving capabilities in ChatGPT are impressive."
* "OpenAI's updates have brought their product up to par with other platforms."
* "I can't wait to try out these new features and see what ChatGPT can do."
Links:
Stack Snacks
OpenAI's DALL·E 3 Announcement: [https://openai.com/dall-e-3] OpenAI's ChatGPT Voice Demo: [https://openai.com/blog/chatgpt-can-now-see-hear-and-speak]
Hosted on Acast. See acast.com/privacy for more information.
Welcome to another episode of Snack Snacks, where we make the complex world of AI digestible! Join John Siwicki at the ungodly hour of 3:45 AM as he dives into the latest updates in the AI landscape. In this episode, we explore:
* Anthropic's Claude Pro: Is the subscription model the future of AI? We discuss the pros and cons of Anthropic's new offering and how it stacks up against other subscription-based AI services.
* Prompting Tips from Anthropic: Learn how to make your prompts more effective with XML tags. A quick guide that could change the way you interact with AI models.
* Repl.it's Ghostwriter: Discover how Repl.it is integrating GPT-4 into their online IDE, and what it means for developers.
* Zoom's AI Features: Zoom is stepping up its game with AI companions in meetings. We delve into the practicality and privacy concerns surrounding this new feature.
* OpenAI's Developer Conference: Mark your calendars for November 6th! We discuss what to expect and why you should be excited.
* Dust, the AA Assistant: A first look at an assistant that combines GPT-4 with your internal company knowledge, connecting platforms like Notion, Google Drive, Slack, and GitHub.
Whether you're an AI enthusiast, a developer, or just curious about the future of technology, this episode has something for you. So grab a snack and tune in!
Key Takeaways:
* Subscription models are becoming increasingly prevalent in the AI space, offering both advantages and drawbacks.
* Effective prompting can enhance your interaction with AI models.
* AI is becoming deeply integrated into everyday tools, from IDEs to video conferencing platforms.
* Upcoming events like OpenAI's Developer Conference are crucial to keep an eye on for the latest innovations.
Don't miss out on this comprehensive rundown. See you in the next episode!
Hosted on Acast. See acast.com/privacy for more information.
John Siwicki is the host of Stack Snacks, a show that aims to make learning about AI simple and fun. He provides insights and commentary on various AI-related topics and shares interesting links and resources with his audience.
Summary:
In this episode of Stack Snacks, John Siwicki discusses several interesting AI-related topics. He starts by introducing Ideogram, an image-to-text generator that handles text in a unique way. He highlights some of the interesting results it produces, such as generating T-shirt designs and novelty items. John also mentions that Ideogram is free to try, but users should be cautious as their prompts are public-facing.
Next, John talks about OpenAI's guide for teachers on how to use GPT in the classroom. He emphasizes that even though it's targeted at teachers, it provides valuable insights and thought exercises for everyone. The guide covers topics like how GPT works, its limitations, and biases. John encourages his audience to check it out and explore different ways of writing prompts.
Moving on, John discusses Loom AI, a new feature added to the screen recording app Loom. He highlights the various enhancements Loom AI offers, such as automatically generating titles, creating summarizations, adding chapters, and providing editing features similar to Descript. John finds the addition of AI-powered talking points and the ability to generate personalized versions of recordings with variables particularly interesting. He mentions that Loom AI is available as an add-on with a monthly fee.
John then introduces Audio Read, a tool that converts articles into voiceovers. He mentions that what sets Audio Read apart is its ability to generate an RSS feed, allowing users to subscribe to the voiceovers in their podcast player. He sees this as a convenient way to consume articles while on the go.
Lastly, John briefly mentions a new Canva plugin called Chat GPT. Although he couldn't find much information about it, he demonstrates how it integrates with Canva and allows users to create logos based on prompts.
Key Takeaways:
* Ideogram is a text-to-image generator that produces high-quality and interesting results.
* OpenAI's guide for teachers on using GPT in the classroom provides valuable insights for everyone, not just teachers.
* Loom AI offers several enhancements to the screen recording app, including automatic title generation, summarizations, chapters, and advanced editing features.
* Audio Read converts articles into voiceovers and provides an RSS feed for easy podcast-style consumption.
* Canva has a new Chat GPT plugin that integrates with the design platform and allows users to create logos based on prompts.
Quotes:
* "Ideogram is a high-quality, free-to-use text-to-image generator that can produce interesting results."
* "OpenAI's guide for teachers on using GPT in the classroom provides valuable insights and thought exercises for everyone."
* "Loom AI offers enhancements like automatic title generation, summarizations, chapters, and advanced editing features."
* "Audio Read converts articles into voiceovers and provides an RSS feed for easy podcast-style consumption."
* "Canva's Chat GPT plugin allows users to create logos based on prompts and integrates seamlessly with the design platform."
Hosted on Acast. See acast.com/privacy for more information.
John Siwicki discusses several interesting topics in this episode. He starts by introducing the Text FX project from Google, which offers AI-powered tools for rappers, writers, and wordsmiths. He explores the different tools available, such as Simile Explode, Unexpected Alliteration, and Acronym Fuse Scene Unfold.
Next, he mentions a project called "Visualizing AI" by Google's DeepMind. This project showcases artwork and animations created by artists based on AI. The visuals are captivating and provide a unique perspective on AI.
John then moves on to discuss a new plugin called Jambot by Figma. Jambot brings Chat GPT inside Figma's whiteboard software, allowing users to utilize AI-generated responses during ideation and collaboration sessions.
He also highlights the launch of Code Llama by Meta Facebook, a model specifically designed for code writing. The model is open-sourced, and Perpexity Labs provides a playground for users to experiment with it.
Lastly, John shares the big news of the week: OpenAI's update to their API, which now allows fine-tuning of Chat GPT Three and a Half Turbo. Fine-tuning enables users to train the model on their own data, resulting in higher quality results, shorter prompts, lower costs, and lower latency requests.
Key Takeaways:
* Google's Text FX project offers AI-powered tools for rappers, writers, and wordsmiths.
* "Visualizing AI" by Google's DeepMind showcases artwork and animations inspired by AI.
* Figma's Jambot plugin brings Chat GPT inside their whiteboard software for collaborative ideation.
* Meta Facebook launched Code Llama, a model tuned for code writing, which is open-sourced and available for experimentation.
* OpenAI's API update allows fine-tuning of Chat GPT Three and a Half Turbo, resulting in higher quality results, lower costs, and lower latency requests.
Links:
* https://www.figma.com/community/widget/1274481464484630971
* https://labs.perplexity.ai/
* https://platform.openai.com/docs/guides/fine-tuning
* https://www.deepmind.com/visualising-ai
* https://textfx.withgoogle.com/
* https://www.youtube.com/@stacksnacks
Hosted on Acast. See acast.com/privacy for more information.
About The Guest(s):
John Siwicki is a tech enthusiast and content creator who shares updates and insights on the latest trends in the tech industry. John provides valuable information and reviews on various platforms and tools.
Summary:
In this episode, John Siwicki shares updates and new developments in the tech industry, particularly in the video and AI space. He discusses the launch of Heygen 2.0, an AI video platform with lifelike avatars, and demonstrates how to create an AI video using Canva. John also highlights the extension of Runway's gen two videos to 18 seconds, allowing for more creative possibilities. He then explores the new features released by Vimeo, including an AI script generator and teleprompter feature. John introduces two content repurposing tools, Decipher AI and Cast Magic, which offer transcription and audiogram generation capabilities. Lastly, he mentions Creerio, a data-driven app with natural language searchability for SQL commands.
Key Takeaways:
* Heygen 2.0 introduces lifelike AI avatars, expanding the possibilities for video creation.
* Runway now supports gen two videos up to 18 seconds, enabling more creative opportunities.
* Vimeo's AI script generator and teleprompter feature simplify video creation and enhance the user experience.
* Decipher AI and Cast Magic offer content repurposing capabilities, including transcription and audiogram generation.
* Creerio provides a user-friendly interface for executing SQL commands with natural language searchability.
Quotes:
* "Heygen 2.0 introduces incredible lifelike AI avatars, revolutionizing video creation."
* "Vimeo's AI script generator and teleprompter feature make video creation a breeze."
* "Decipher AI and Cast Magic offer powerful tools for content repurposing and transcription."
* "Creerio simplifies data analysis with its natural language searchability for SQL commands."
- https://www.deciphr.ai/
- https://vimeo.com/features/ai-script-generator -
https://twitter.com/joshua_xu_/status/1689019874667024384?s=20 - - https://twitter.com/runwayml/status/1689630007746764803?s=20
- https://www.deciphr.ai/
- https://www.castmagic.io
- https://www.querio.ai
Hosted on Acast. See acast.com/privacy for more information.
In this episode, John Siwicki dives into the latest updates from the AI world. He starts with six new features rolling out for ChatGPT, including prompt examples, suggested replies, and the default use of ChatGPT-4 for Plus subscribers. He also discusses the new ability to upload multiple files for the code interpreter and the introduction of keyboard shortcuts.
John then moves on to Google's AI-powered search experience, which has received three significant updates: the inclusion of more videos and images in search results, increased speed, and the addition of cited sources and links. He also touches on Perplexity AI's new file upload feature and how it can quickly summarize PDFs.
Finally, he explores Meta's AudioCraft, a tool for sound generation, and its impressive examples. From sirens to the sound of typing on a typewriter, AudioCraft is pushing the boundaries of AI-generated audio.
Tune in for a comprehensive look at these updates and how they're shaping the AI landscape.
Links:
https://help.openai.com/en/articles/6825453-chatgpt-release-notes
https://audiocraft.metademolab.com/audiogen.html
https://blog.google/products/search/google-search-generative-ai-august-update/
Hosted on Acast. See acast.com/privacy for more information.
Welcome back to another episode with your host, John Swicki. This week, we're diving into some exciting new releases from AWS, learning how to train a llama for your own purposes, and exploring a few other intriguing topics.
AWS's new release, HealthScribe, is our first point of discussion. This innovative tool automatically generates clinical notes from patient and clinical conversations, enhancing clinical productivity with AI-generated notes that are easy to reference, edit, and finalize.
We delve into the implications of HealthScribe, discussing its potential impact on medical workflows and patient consultations.
Next up, we found a fascinating blog post detailing how to fine-tune your own LlamaTube model, Facebook's open-source project.
We walk through the Python-heavy tutorial, discussing how you can use this model for your own purposes.
We then turn our attention to LangChain, a tool designed to assist developers in building applications they previously could not.
We discuss the potential of LangChain, highlighting its ability to combine different sources of computation and knowledge.
Finally, we discuss a thought-provoking blog post titled "AI and the Decline of the Open Web." We explore the implications of login walls, the transformation from forums to private groups, and the rise of walled gardens.
We delve into the potential fragmentation of data and the open web, considering the future of the advertisement model and the increasing value of data sets.
Links: https://aws.amazon.com/healthscribe/ https://mlabonne.github.io/blog/posts/Fine_Tune_Your_Own_Llama_2_Model_in_a_Colab_Notebook.html https://blog.langchain.dev/automating-web-research/ https://tanay.substack.com/p/ai-and-the-decline-of-the-open-web
https://www.youtube.com/@stacksnacks
Hosted on Acast. See acast.com/privacy for more information.
In this episode, host John Swicki takes a deep dive into the current state of AI, discussing newly released open-source models, on-device applications, and possible future developments from major tech companies.
Highlights of this episode include:
* Llama 2's Arrival: John kicks off the episode by reviewing Facebook Meta's recently open-sourced AI model, Llama 2. Discussing its complexity, sizes (7 billion, 13 billion, and 70 billion parameters), and potential for both research and commercial use, John predicts interesting developments as more developers gain access to this tool.
* Qualcomm's Ambition: Next, the focus shifts to Qualcomm's plan to run Llama 2 on devices. John explores the potential for increased personalization, better security, and the avoidance of cloud infrastructure reliance.
* OpenAI's ChatGPT Custom Instructions: Then, John breaks down OpenAI's new feature - 'Custom Instructions' for ChatGPT. This enhancement allows users to provide ChatGPT with personalized information for improved response quality, changing the way we interact with AI.
* Apple's AI Potential: To conclude the episode, John speculates about Apple's rumored AI project, internally referred to as Apple GPT. How might this affect Siri and Apple's privacy-centric approach? Stay tuned to find out.
Links mentioned in this episode:
* Perplexity Labs' Llama Playground: https://labs.perplexity.ai/
* Stack Snacks Youtube: https://www.youtube.com/channel/UC-lzS2PWvUztxHIMdf55S5Q
Join John next time for more insights into the world of artificial intelligence, and don't forget to subscribe to our podcast so you never miss an episode!
Hosted on Acast. See acast.com/privacy for more information.
Welcome to another episode where we dive deep into the exciting world of artificial intelligence. Here's what we cover in this episode:
* OpenAI-AP Collaboration: We discuss the new agreement between OpenAI and the Associated Press to share select news content and technology. OpenAI will have access to part of AP's text archive to refine its model, and this collaboration could possibly mark the start of a new era in news services.
* Claude 2 Update: Anthropic recently launched Claude 2, a significant update from its predecessor. We explore the features of this AI model and its user-friendly interface, accessible now at claude.ai. We also delve into its performance and its newly-introduced file upload feature.
* Shopify's Sidekick: Shopify is set to release a new feature - a co-pilot named Sidekick, aimed at aiding users in setting up their Shopify account. We touch upon its features and potential impact on e-commerce business owners.
* Poe.com Update: Poe.com has introduced some significant updates, including file uploading across all their offerings and even processing URLs to enhance user experience. We also discuss their two new OpenAI models with larger context windows and the introduction of their Mac app.
* Google's SoundStorm: We explore Google's newly-published research paper 'SoundStorm', on audio generation and what it could possibly mean for the future of AI.
* Other Updates: We discuss some quick updates about Bard, another AI tool, and the link to a separate video covering its updates in more detail.
Links to all topics covered, as well as some related resources, are included in the podcast description. We encourage listeners to check them out for a deeper understanding of each subject.
Thank you for joining us in this episode. We appreciate your interest and look forward to bringing you more updates from the fascinating world of AI!
Links:
https://www.anthropic.com/index/claude-2
https://www.ap.org/press-releases/2023/ap-open-ai-agree-to-share-select-news-content-and-technology-in-new-collaboration
https://twitter.com/poe_platform/status/1678843136452468754 https://twitter.com/Shopify/status/1679119634685984768
https://ai.googleblog.com/2023/06/soundstorm-efficient-parallel-audio.html
Bard Video
Hosted on Acast. See acast.com/privacy for more information.
From the publisher's feed
Explore the latest in tech each week with host John Siwicki on the Stack Snacks podcast. This snack-sized audio experience makes cutting-edge AI, product design, and development digestible. Get a…
https://www.youtube.com/@stacksnacks
www.stack-snacks.com
Hosted on Acast. See acast.com/privacy for more information.