Another Crazy Day in AI 3/63

Another Crazy Day in AI: Class is in Session with Claude

Apr 8
3 min read

Hello, AI Enthusiasts.

You made it through Tuesday. Now here’s what AI did while you worked.

Anthropic just launched a new initiative for colleges and universities. This isn’t about quick answers—it’s about helping students think critically and making life easier for educators. Who says AI can’t hit the books?

Meanwhile, a crypto founder turned entrepreneur breaks down how small business owners are using AI to do more with less. Startups playing like the big kids? It’s already happening.

Microsoft’s latest update brings AI-powered call summaries to Dynamics 365. What does it mean when even your voicemails get smarter?

AI’s here to lighten your load—let’s get creative.

Here's another crazy day in AI:

New campus tools from Anthropic
Why smart entrepreneurs let AI take the lead
Microsoft adds AI call summaries to Dynamics 365
Some AI tools to try out

TODAY'S FEATURED ITEM: Claude for Education—Built for Higher Ed

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 happens when a tech company builds an AI experience specifically for the unique pace and purpose of university life?

Anthropic has launched Claude for Education, a version of its AI platform developed specifically for higher education. It’s designed to support how students learn, how faculty teach, and how administrators manage academic life—all within a secure, reliable system. This new offering reflects an ongoing shift: instead of seeing AI as an add-on, many colleges and universities are starting to consider how it can become part of their everyday academic framework. The announcement post outlines the initial rollout, partnerships with major institutions, and the new features designed to work across the entire academic community.

Here are some important points about this initiative:

Claude for Education is tailored for students, faculty, and academic staff, addressing their specific needs.
The platform includes safeguards to promote academic integrity and responsible usage.
Transparency and data privacy are prioritized, which are essential in educational contexts.
Anthropic is collaborating with institutions like the University of California, the University of Pennsylvania, and NYU to gather feedback and refine the platform.
New classroom-specific prompts and policies have been developed to reflect real academic challenges.

The introduction of Claude for Education offers a promising path toward enhancing the learning experience in higher education. By focusing on the distinct needs of students and educators, this initiative aims to create a supportive environment where technology acts as a helpful partner in the learning process. The potential for fostering deeper engagement and collaboration is significant, allowing both students and faculty to benefit from an AI that understands their unique challenges.

As we consider the integration of AI into academic settings, we might ask ourselves: how do we strike the right balance between leveraging technology and maintaining the human connections that make education so impactful? Claude for Education encourages us to reflect on this balance, inviting us to explore innovative ways of teaching and learning that enhance the overall educational experience. Ultimately, the goal is to cultivate an academic culture where technology enriches human interaction, allowing everyone—students, educators, and administrators—to thrive in a supportive, dynamic learning environment.

Read the announcement here.

OTHER INTERESTING AI HIGHLIGHTS:

Why Smart Entrepreneurs Let AI Take The Lead

/Vladimir Gorbunov on Entrepreneur Media

In Entrepreneur Media, Vladimir Gorbunov outlines how founders and small business owners are using AI to do more with less — automating operations, improving sales, and delivering enterprise-level customer support on a startup budget. Tools like Zapier, Notion AI, Jasper, and Intercom are helping lean teams scale faster, boost productivity, and focus on high-impact work. Gorbunov shares real tools and actionable advice for anyone looking to turn AI into a competitive edge. His biggest tip? Start small, and let AI handle the busywork.

Read more here.

Microsoft Adds AI Call Summaries To Dynamics 365

/Isabel Greene, Microsoft Dynamics 365 Blog

Microsoft’s latest update to Dynamics 365 Contact Center adds an AI-powered summary feature that helps reps manage call quality and customer experience more efficiently. Service agents can now view detailed AI summaries, sentiment analysis, and post-call surveys — plus full transcripts and multi-session timelines for every customer conversation. Whether it’s a voicemail or a live support call, AI now handles the wrap-up. Admins can enable these summaries directly in the Copilot Service admin center.

Read more here.

SOME AI TOOLS TO TRY OUT:

Writehere.site – Open framework for natural, long-form AI writing.
EverTutor Live – Voice-based AI tutor that adapts to how you learn.
Kozy – Create pro-level edited videos from simple prompts.

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!!!

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: Better Ads by Prompt

Apr 7
4 min read

Hello, AI Enthusiasts.

Welcome to another week! If you thought AI would take a break over the weekend, think again—it’s been busy...

Ethan Mollick shares a study revealing generative AI often surpasses human creativity in marketing tasks. The upcoming paper compares AI-generated visuals to human designs. Who comes out on top?

In healthcare news, an AI-assisted study found that AI-generated diagnoses were rated equal or better than those from physicians. Who knew AI could play doctor too?

And Google’s AI Mode now supports image queries—time to snap and ask your way to knowledge!

Get ready for a rollercoaster week in AI.

Here's another crazy day in AI:

Can a prompt outperform a pro?
Can AI make better first calls in urgent care?
Google AI Mode now answers image-based questions
Some AI tools to try out

TODAY'S FEATURED ITEM: When AI Creates Better Ads

Image Credit: Wowza (created with Ideogram)

Is your next best-performing ad just a prompt away?

That’s the idea behind The Power of Generative Marketing, a forthcoming academic study by Jochen Hartmann, Yannick Exner, and Samuel Domdey, soon to appear in the International Journal of Research in Marketing. The paper explores how generative AI compares with professional human creatives when it comes to producing marketing visuals—and whether it can, in some cases, do even better.

The research team analyzed thousands of images generated using tools like DALL·E 3, Midjourney v6, and Adobe Firefly 2. These AI visuals were tested against professional stock photography and content designed by freelancers. Using controlled experiments, perception studies, and a real-world ad campaign, the researchers measured both the creative quality and the business results.

Wharton professor Ethan Mollick shared the study in a recent post, calling attention to how well generative AI performed. He noted that these tools aren’t just closing the gap with human creators—they’re starting to outperform them in specific tasks. “Generative tools are already producing superhuman performance for many marketing tasks,” he wrote. What’s striking, Mollick added, is that these aren’t projections. This is already happening.

Hartmann, Jochen and Exner, Yannick and Domdey, Samuel, The power of generative marketing: Can generative AI create superhuman visual marketing content? (September 05, 2024). International Journal of Research in Marketing, Forthcoming., Available at SSRN: https://ssrn.com/abstract=4597899 or http://dx.doi.org/10.2139/ssrn.4597899 — Hartmann, Jochen and Exner, Yannick and Domdey, Samuel, The power of generative marketing: Can generative AI create superhuman visual marketing content? (September 05, 2024). *International Journal of Research in Marketing*, Forthcoming., Available at SSRN: https://ssrn.com/abstract=4597899 or http://dx.doi.org/10.2139/ssrn.4597899

What the study revealed about AI-generated marketing content:

AI visuals scored higher in visual quality, appeal, and how well they fit the creative brief.
In a live campaign involving over 173,000 participants, an image made with DALL·E 3 received up to 50% more clicks than the professional-designed version.
The cost of generating the top-performing image was just $0.04.
Even without knowing which was AI-generated, human judges consistently preferred the synthetic visuals.
The team also released GenImageNet, a new dataset to support further research on generative marketing.

The findings paint a clearer picture of where we are with AI in creative work. In marketing, visuals need to do more than look good—they have to communicate quickly, align with brand identity, and drive results. This study suggests that generative tools are starting to meet, and sometimes exceed, those expectations in controlled settings and real-world campaigns.

Of course, these results don’t mean AI is “replacing” human creativity. What they do suggest is that the creative process itself is evolving. Tools like DALL·E and Midjourney are giving marketers new ways to test, iterate, and experiment—often in a matter of minutes and at a fraction of the cost. For teams stretched thin or looking to scale content quickly, this opens up new possibilities. Still, strategy, storytelling, and brand context remain essential, and human oversight is what gives these tools their impact.

This research is one of the clearest signals yet that generative AI is becoming part of how creative work gets done—not as a gimmick, but as a tool that performs under pressure. As more studies like this emerge, the conversation shifts from if we use AI in creative tasks to how we integrate it meaningfully.

Read Mollick's LinkedIn post here.

Read and download the full paper here.

OTHER INTERESTING AI HIGHLIGHTS:

Can AI Make Better First Calls in Urgent Care?

/Dan Zeltzer, Zehavi Kugler, Lior Hayat, Tamar Brufman, et al. published on Annals of Internal Medicine

In a study of AI-assisted virtual urgent care visits, researchers compared AI-generated diagnoses and treatment recommendations with those of physicians. Surprisingly, AI recommendations were rated of equal or better quality in most cases—especially when it came to identifying critical red flags and adhering to clinical guidelines. While doctors excelled at adapting to evolving information during a consultation, AI held its own in initial decision-making. The results suggest AI could become a powerful decision-support partner in virtual care settings.

Read more here.

Google AI Mode Now Answers Image-Based Questions

/Barry Schwartz on Search Engine Land

Google is rolling out an upgraded AI Mode that now supports multimodal inputs—you can ask questions using images from your camera or uploads. Drawing on its experience with visual search and Google Lens, AI Mode interprets scenes, identifies objects, and understands context to generate richly informed responses. This update expands access to millions more users and hints at a future where search is no longer just typed text, but interactive and visual. Try snapping a photo and asking away.

Read more here.

SOME AI TOOLS TO TRY OUT:

Clockwise – Smart scheduling and calendar automation with AI.
Experiments – Turn personal challenges into loggable, trackable experiments.
Enconvo – Build custom workflows that connect all your favorite tools.

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!!!

Another Crazy Day in AI: How DeepMind Plans to Keep AGI in Check

Apr 4
3 min read

Hello, AI Enthusiasts.

Before you switch to weekend mode, here’s a quick AI rundown:

DeepMind is getting serious about AGI safety. In a new paper, their top researchers explore the future of human-level AI—and how we might avoid its biggest pitfalls. Think: solving diseases vs. losing control over machines. Heavy stuff, but important reads.

Meanwhile, AI agents could soon be weaponized—cybersecurity experts are already tracking their early moves.

In a creative twist, Stanford brought filmmakers and AI researchers together to explore how narratives influence our understanding of artificial intelligence.

Now, kick back and relax—you’ve earned it!

Here's another crazy day in AI:

The blueprint for responsible AGI
Is the future of hacking agentic too?
Stanford workshop on AI and storytelling
Some AI tools to try out

TODAY'S FEATURED ITEM: Google DeepMind’s AGI Plan

Image Credit: Wowza (created with Ideogram)

What happens when general-purpose AI reaches a level of capability that rivals—or even surpasses—our own?

In a new paper An Approach to Technical AGI Safety and Security, a team from Google DeepMind—including Anca Dragan, Rohin Shah, Four Flynn, and Shane Legg—shares their latest thinking on how to responsibly navigate the development of artificial general intelligence (AGI), a kind of AI that could match or exceed human cognitive abilities across most tasks. This piece outlines the risks AGI presents and the robust systems DeepMind is building to manage those risks.

The authors explore the balance between optimism for AGI’s benefits and deep caution about its potential harms. From improved diagnostics and personalized learning to cybersecurity and misalignment with human values, they break down where things could go wrong—and what they’re doing to make sure they don’t.

A few things they’re working on:

Mapping out four central risk areas: misuse, misalignment, accidents, and broader societal impacts
Building technical safeguards like access restrictions and scenario simulations
Refining training methods that include human feedback and uncertainty-aware behavior
Investing in interpretability tools (such as MONA) to understand how AI systems reach their decisions
Conducting regular safety evaluations and inviting independent input
Stress-testing systems early and often to adapt as the technology evolves

The paper doesn’t offer a silver bullet—and it doesn’t try to. Instead, it lays out a living framework that’s meant to grow and shift alongside the development of AGI itself. The goal is not just to anticipate what might go wrong, but to put systems in place that are capable of responding when the unexpected does happen.

For those tracking how AGI is being shaped behind the scenes, this paper provides a window into how one of the leading research labs is thinking about responsibility at scale. It doesn’t shy away from the complexity of the challenge, nor does it overstate what’s been solved. It simply asks: how do we build with care, when the stakes are this high?

Read the full article here.

Read the full paper here.

OTHER INTERESTING AI HIGHLIGHTS:

Is the Future of Hacking Agentic Too?

/Rhiannon Williams on MIT Technology Review

AI agents are becoming smarter and more autonomous—and cybersecurity experts warn that they might soon be used to conduct cyberattacks at scale. Unlike basic bots, agents can adapt, plan, and execute attacks with alarming efficiency, posing a serious threat to digital infrastructure. A project called LLM Agent Honeypot is already working to track these AI-driven intrusions in real time. Researchers say it's only a matter of time before cybercriminals start relying on agents for hacking—and we need to be ready before it happens.

Read more here.

Stanford Workshop on AI and Storytelling

/Dylan Walsh on Stanford HAI News

What happens when filmmakers and AI researchers work together to tell stories about artificial intelligence? A workshop at Stanford’s HAI brought both groups together to explore how narratives shape public understanding—and policy—around AI. Participants like filmmaker Sophie Barthes and researcher John Thickstun shared how blending academic ideas with storytelling revealed just how hard (and important) it is to make complex tech human, accessible, and emotionally compelling. The initiative is a unique reminder that how we talk about AI may shape how we use it.

Read more here.

John Thickstun, an assistant professor at Cornell, and filmmaker Sophie Barthes collaborate on a screenplay. | Source: Stanford HAI News — John Thickstun, an assistant professor at Cornell, and filmmaker Sophie Barthes collaborate on a screenplay. | Source: *Stanford HAI News*

SOME AI TOOLS TO TRY OUT:

Beautiful AI – Instantly design stunning presentations with AI.
Cove Apps – A visual workspace where you can build custom AI-powered interfaces.
ElevenLabs – Just launched a text-to-bark model for dogs.

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!!!