top of page
Another Crazy Day in AI: An Almost Daily Newsletter

Hello, AI Enthusiasts.


Whew—did this week fly by, or was that just the caffeine?


If you think you’ve seen it all, wait till you hear about a new way students are learning—using nothing but the air around them. In Taiwan, researchers built a virtual writing system where students can draw or write midair. No screen. No stylus. Just gestures.


Meanwhile, AI pioneers are now training AI to catch other AIs—before they mislead, deceive, or go rogue.


And just when we thought HeyGen was done, Avatar IV drops a major upgrade. Wanna know what’s new?


Here's another crazy day in AI:

  • A new classroom tool that tracks finger movement

  • AI pioneer launches nonprofit for honest intelligence

  • HeyGen levels up with hyper-real Avatar IV

  • Some AI tools to try out


TODAY'S FEATURED ITEM: Bringing Motion-Based Input to Schools


A robotic scientist in a classic white coat with 'AI Scientist' on its back stands beside a human scientist with 'Human Scientist' on their coat, looking towards the AI Scientist.

Image Credit: Wowza (created with Ideogram)


What if students could write in midair—no paper, no pens—and still be fully engaged in class?


A team of researchers in Taiwan has developed technology that might change how we think about traditional classroom activities. Led by Associate Professor Liang-Bi Chen at the National Penghu University of Science and Technology, the research group created a virtual writing platform where students can create handwriting and artwork by simply moving their fingers through empty space. Published in the IEEE Canadian Journal of Electrical and Computer Engineering, this work tackles real problems teachers face every day while opening up new possibilities for student engagement.


In a new article for IEEE Journal Watch, freelance writer Michelle Hampson takes a closer look at how the system works. It relies on wearable sensors that track finger movement, converting gestures into written characters or sketches on screen. Unlike other classroom tools that depend on physical devices like styluses or tablets, this setup is designed for direct interaction without contact. To build it, the researchers used a microcontroller and a gyroscope module—components selected with cost and classroom practicality in mind.



Here’s what the researchers focused on:

  • How accurately the system could replicate handwritten characters drawn in the air

  • The usability of the interface from a student’s perspective

  • Whether students found the experience intuitive or difficult to adapt to

  • The kinds of improvements that might be needed before use in real classrooms


There’s still a lot to figure out before something like this becomes common in classrooms. Questions around usability, classroom management, and even how students respond to this kind of interaction will take time to answer. But ideas like this are a reminder that innovation in education doesn’t always have to mean high-end, high-cost equipment. Sometimes it’s about finding new ways to make familiar practices—like handwriting—more flexible and accessible.


It also opens up room to think differently about participation. Not every student is equally comfortable using traditional tools, and not every classroom has the same resources. A system that lowers barriers while still encouraging creative expression could offer real value. It’s early, but the work by Chen’s team gives educators and researchers something concrete to consider—and maybe build on.




Read the full article here. Read the study here.

OTHER INTERESTING AI HIGHLIGHTS:


AI Pioneer Launches Nonprofit For Honest Intelligence

/Dan Milmo, Global Technology Editor, on The Guardian


Yoshua Bengio, a Turing Award-winning computer scientist known as one of the “godfathers” of AI, has launched LawZero, a nonprofit focused on creating “honest” AI. Its core project, Scientist AI, aims to predict and prevent harmful or deceptive behavior by autonomous AI agents—acting more like a psychologist than an assistant. Unlike typical generative AI, it won’t give definitive answers but will instead offer probabilities, building a system that’s transparent and humble. Bengio stresses the importance of developing a watchdog AI that is just as advanced as the agents it monitors.



Read more here.


HeyGen Levels Up with Hyper-Real Avatar IV

/HeyGen


HeyGen has rolled out a major update to Avatar IV, enhancing its avatars with lifelike gestures, micro-expressions, and emotional delivery—all prompt-controlled. The upgrade allows creators to direct tone, posture, and facial reactions with simple cues like “casual tone” or “spread hands,” resulting in avatars that truly perform. Videos can now be generated in full HD (1080p) for up to 60 seconds, with extended options for enterprise users. With natural motion and expressive nuance, Avatar IV brings digital characters closer than ever to human realism.



Read more here.

SOME AI TOOLS TO TRY OUT:


  • Jots – Boost personal growth with an AI-powered journaling app.

  • Onda – Turn podcasts into searchable, timestamped notes.

  • Picbolt – Make your screenshots pop with standout visuals.


That’s a wrap on today’s Almost Daily craziness.


Catch us almost every day—almost! 😉

EXCITING NEWS:

The Another Crazy Day in AI newsletter is on LinkedIn!!!



Wowza, Inc.

Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.





Another Crazy Day in AI: An Almost Daily Newsletter

Hello, AI Enthusiasts.


Winding down the week, but here’s something to wake your brain before the weekend hits:


HubSpot just dropped something big—its new CRM deep research connector with ChatGPT. That means businesses can now talk to their data in plain English. No code. No data team. Just questions, answers, and faster decisions.


Meanwhile, Google’s still hiring engineers, even in the age of AI. Pichai says we’re nowhere near done needing humans to build this future.


And in the world of science, researchers are split on where AI belongs in peer review and academic writing. Helpful? Maybe. Acceptable? Depends who you ask.


Keep these AI insights handy as you step into the weekend.


Here's another crazy day in AI:

  • ChatGPT now works with HubSpot

  • Alphabet CEO plans to keep hiring engineers

  • Survey finds ethical divide over AI use in papers

  • Some AI tools to try out


TODAY'S FEATURED ITEM: HubSpot Adds AI Research


A robotic scientist in a classic white coat with 'AI Scientist' on its back stands beside a human scientist with 'Human Scientist' on their coat, looking towards the AI Scientist.

Image Credit: Wowza (created with Ideogram)


What if your CRM could answer the kinds of questions you'd normally ask a data team?


Karen Ng, SVP of Product at HubSpot, recently shared in a LinkedIn post a new integration that brings ChatGPT’s Deep Research tool directly into the HubSpot ecosystem. It’s the first CRM to offer this kind of connection, making it possible for users to interact with their customer data using natural language—without needing advanced analytics skills or coding experience. The goal? To help small and mid-sized teams do the kind of strategic analysis that once required larger departments and more technical expertise.


The connector lets teams pull insights from both structured and unstructured customer data—everything from contact records to ticket sentiment. Once connected, you can simply ask questions in plain language and get actionable answers that can help guide decision-making across marketing, sales, support, and customer success teams.



What the integration makes possible:

  • Ask direct questions about your customer base—no queries, just plain language

  • Surface patterns around lead quality, churn signals, or campaign performance

  • Explore opportunities by filtering data through context like industry, tech use, or team size

  • Equip support and success teams to anticipate volume, track sentiment, and plan outreach

  • Keep access levels secure—only users with existing permissions can retrieve relevant insights

  • Integrate the tool without disrupting workflows; once a Super Admin enables it, it’s ready for use


The integration is available to HubSpot users with any paid ChatGPT plan—Plus, Pro, Team, Enterprise, or Edu (with some differences for EU-based accounts). HubSpot confirmed that the data is not used for model training and that permission rules stay intact, helping teams maintain control of sensitive information.



While it doesn’t replace more traditional data analysis methods, the integration does expand the kind of questions everyday users can explore without needing to involve a data team. The real benefit may come in those early stages of inquiry—when someone is trying to understand what’s happening, explore a trend, or test a hunch—and now has a more direct path to insight.


It also invites a broader range of team members into the conversation about what the data might suggest. For organizations already working within HubSpot, this may be an opportunity to use what's already available more effectively. And for others watching from the outside, it raises some thoughtful considerations: how much of our existing data sits underused, simply because accessing it has always felt just out of reach?




Read the full post here.

OTHER INTERESTING AI HIGHLIGHTS:


Alphabet CEO Plans To Keep Hiring Engineers

/Julia Love, Tech Reporter at Bloomberg, featured in The Japan Times


Alphabet CEO Sundar Pichai says Google will keep hiring engineers through at least 2026, even as AI advances accelerate. Speaking at the Bloomberg Tech conference, Pichai emphasized the need for human talent to unlock the full potential of AI, especially as it transforms engineering productivity. While optimistic about AI’s capabilities, he acknowledged the models still make basic mistakes and that general artificial intelligence (AGI) isn’t guaranteed. Pichai also addressed concerns from publishers about AI-generated search overviews, promising that Google remains committed to supporting web traffic.



Read more here.


Survey Finds Ethical Divide Over AI Use In Papers

/Diana Kwon, Freelance Science Journalist, on Nature


A global survey of 5,000 researchers published in Nature reveals deep divisions over when and how AI should be used in scientific writing and peer review. While most agree that using AI to edit or translate is acceptable, opinions vary on whether disclosure is necessary and whether AI should draft papers or reviews. The data shows that actual use of AI tools remains low, especially for tasks like peer review or generating first drafts. The survey also found differences based on researchers’ career stages and native languages, suggesting that younger and non-native English-speaking academics are more open to AI assistance.



Read more here.

Image Credits: Nature
Image Credits: Nature

SOME AI TOOLS TO TRY OUT:


  • Spine AI – Creates editable research reports from thousands of sources.

  • Amy – AI agent for private market investors that handles the busywork.

  • SlidesAI – Instantly generate beautiful presentations right inside Google Slides.


That’s a wrap on today’s Almost Daily craziness.


Catch us almost every day—almost! 😉

EXCITING NEWS:

The Another Crazy Day in AI newsletter is on LinkedIn!!!



Wowza, Inc.

Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.





Another Crazy Day in AI: An Almost Daily Newsletter

Hello, AI Enthusiasts.


Midweek mood check: holding it together, or barely holding your charger?


While you're catching your breath, HeyGen just dropped a new editor that lets you make polished videos with nothing but a script! Creativity just got a serious upgrade, with you in full control.


Speaking of control, Anthropic’s testing the waters with AI-powered blog posts, polished by human hands. But... the water’s getting choppy—Reddit’s suing them over data use without permission.


Here's another crazy day in AI:

  • HeyGen introduces AI Studio

  • Anthropic gives Claude a blog

  • Reddit sues Anthropic over data use

  • Some AI tools to try out


TODAY'S FEATURED ITEM: Inside HeyGen AI Studio


A robotic scientist in a classic white coat with 'AI Scientist' on its back stands beside a human scientist with 'Human Scientist' on their coat, looking towards the AI Scientist.

Image Credit: Wowza (created with Ideogram)


What if creating professional videos became as simple and intuitive as writing a document?


HeyGen recently introduced their AI Studio during the company's first-ever keynote, presenting a reimagined way to create videos—especially for those who don't typically work with editing software. The platform brings together tools like avatar-driven A-roll, customizable voices, and gesture cues, all organized within a document-style interface designed to make the process more approachable and collaborative. It's a notable move toward simplifying video production without stripping it of nuance or creative control.



Here’s what stands out from the launch:

  • AI Studio offers an intuitive editor that moves away from traditional timelines, making it easier for creators of all levels to produce video content.

  • New A-roll capabilities push the realism of avatars in appearance, movement, and vocal nuance—bridging the gap between synthetic and human-like delivery.

  • Voice Director Mode allows users to shape tone and pacing using simple prompts, much like directing a live actor.

  • Voice Mirroring enables users to upload short recordings of their own voice to guide how their avatar speaks—down to rhythm and emotion.

  • Gesture Control lets creators link specific words or phrases to gestures, allowing for more deliberate communication.

  • A built-in collaboration environment supports tagging, multi-user editing, previews, proofreading tools, and visual templates.

  • Upcoming features promise more control over B-roll with options like camera movement, motion elements, and generative visuals guided by text prompts.



The announcement comes at a time when video content has become essential for most businesses, yet many still struggle with the technical barriers of traditional editing software. HeyGen's approach addresses this gap by treating video creation more like document editing—a familiar process for most professionals. The company serves over 15 million users across 179 countries, suggesting there's substantial demand for more accessible video tools. CEO Joshua Xu framed the launch around moving beyond "good enough" results, though the practical impact will depend on how well these tools perform in everyday use cases.


Watching this development unfold raises questions that go beyond the technology itself. We're seeing more people who never considered themselves video creators suddenly having access to professional-grade tools. A marketing manager who's comfortable writing emails might now produce weekly video updates. A small business owner could create training materials without hiring a production team. But this accessibility also means the bar for what constitutes compelling video content keeps rising. When everyone has access to polished production tools, standing out requires more than just technical proficiency—it demands genuine insight, clear thinking, and authentic communication. The real test for platforms like HeyGen won't be whether they can generate professional-looking videos, but whether they help people say something worth watching.




Read the full article here.

Watch the full recording here.

OTHER INTERESTING AI HIGHLIGHTS:


Anthropic Gives Claude a Blog

/Kyle Wiggers, AI Editor, on TechCrunch


Anthropic has quietly launched Claude Explains, a blog featuring AI-generated posts written by its Claude model and refined by human editors. The blog showcases how AI and subject matter experts can co-create technical content on topics ranging from coding to business strategy. While the homepage implies Claude writes everything, Anthropic clarifies that each post undergoes editorial oversight to add insights and ensure accuracy. This move reflects a growing trend among tech companies experimenting with AI-generated content, even as others face public backlash over AI-written material riddled with errors.



Read more here.


Reddit Sues Anthropic Over Data Use

/Ashley Capoot, Reporter, and Jonathan Vanian, Tech Reporter, on CNBC


Reddit is suing Anthropic, accusing the AI startup of violating its terms by using Reddit’s data without authorization to train its models. The lawsuit alleges "unlawful and unfair" practices and contrasts Anthropic's actions with OpenAI and Google, which both license Reddit content. The complaint frames Anthropic as disregarding the platform’s rules while profiting from its user-generated communities. Reddit seeks damages and wants Anthropic to comply with contractual obligations, escalating the ongoing battle over how AI companies source their training data.



Read more here.

SOME AI TOOLS TO TRY OUT:


  • Blackbox – Ask anything, build apps, do deep research, and generate images—all with AI.

  • ShumerPrompt – A clean, well-designed AI prompt library by Matt Shumer.

  • Chance – Snap photos and get instant visual explanations using AI.


That’s a wrap on today’s Almost Daily craziness.


Catch us almost every day—almost! 😉

EXCITING NEWS:

The Another Crazy Day in AI newsletter is on LinkedIn!!!



Wowza, Inc.

Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.





Copyright Wowza, inc 2025
bottom of page