AI DUNN Right Weekly - Issue #26
- Mar 23
- 6 min read
Practical AI insights for business growth
Hey AI Innovators! đź‘‹
This week had a story that made me put my phone down and stare at the ceiling. A regular person with no medical training used ChatGPT to create a cancer vaccine for his dog. And it worked. I'll get to that. But first... something that's going to affect your wallet.
Here's what you need to know:
Sam Altman wants to put AI on a meter, and every business owner needs to hear this
A man used ChatGPT to create a cancer vaccine for his dog. The tumour shrank by half.
OpenAI called Anthropic's enterprise dominance a "wake-up call" in a company-wide meeting
Anthropic interviewed 81,000 people about AI... and the findings are complicated
Val Kilmer will star in a new film, one year after his death
Read time: 7 minutes
Big Story
Sam Altman Wants to Put AI on a Meter
Right now, most of us pay a flat monthly fee for our AI tools. Twenty dollars, thirty, maybe fifty... and we use as much as we want. Priced like Netflix, not electricity.
Sam Altman just told us that probably won't last.
His vision for AI pricing? Like a utility bill. You use it, you pay for what you consume, you get an invoice at the end of the month. Intelligence on a meter.
Here's what that means for your business:
Local AI tools that run on your own device are getting genuinely good, fast
The gap with frontier models like ChatGPT and Claude is closing every month
If metered pricing arrives, businesses with no AI strategy will feel it first
For now, nothing changes if you're on a flat $20 to $30 plan. But the businesses that win will be the ones treating AI as a tracked, intentional operating expense.
My take on this? Start identifying which AI tasks are genuinely high-value for your business. Because one day, those are the ones worth paying more for. The low-value habits... you'll want to drop before the meter is running.
Source: Mindstream
What's New This Week
A Man Used ChatGPT to Build a Cancer Vaccine for His Dog. It Worked.
Paul Conyngham is an Australian with a computer science background. Not a doctor. Not a researcher. His dog Rosie had a tennis-ball-sized tumour, so he got to work:
Asked ChatGPT how to approach creating a personalised mRNA cancer vaccine
Got Rosie's DNA sequenced at a university lab for $3,000
Used DeepMind's AlphaFold to identify the mutated proteins
Had the vaccine manufactured and injected it
The tumour shrank by half. One researcher who reviewed the work said "I was like, holy crap. It worked." File this one away the next time someone tells you AI is only going to cause harm.
Source: AI with Kyle
OpenAI Is in "Wake-Up Call" Mode Over Anthropic's Enterprise Lead
Fidji Simo, OpenAI's CEO of Applications, told staff in a company-wide meeting that Anthropic's dominance in enterprise is a "wake-up call." Her exact words: OpenAI "cannot miss this moment because we are distracted by side quests."
While OpenAI was building browsers and exploring hardware, Anthropic quietly became the go-to AI for businesses. OpenAI is now pulling back hard and refocusing on:
Coding tools (Codex hit 2 million weekly users since January)
Business customers and enterprise deals
For you? This competition is great news. Both tools will get better and more business-focused because of it.
Source: The Rundown AI
Anthropic Asked 81,000 People How They Really Feel About AI
Anthropic used Claude itself to interview 81,508 people across 159 countries in 70 languages... all in a single week. The results were more nuanced than the usual "people hate AI" headlines:
Top hope: professional excellence, using AI to do better work and free up time
Top fear: AI getting things wrong, followed closely by job anxiety
Sentiment ran above average in India and South America, neutral to negative in the US, Europe, Japan, and South Korea
Most people aren't choosing between hope and fear. They're holding both at the same time.
Source: The Rundown AI
Val Kilmer Will Star in a New Film, One Year After His Death
The film As Deep As The Grave will use generative AI to bring Kilmer's performance to life. He had been cast five years ago but illness prevented him from filming. His estate and daughter worked closely with the production on it.
This isn't completely out of nowhere. Kilmer had already used AI voice tech after throat cancer damaged his voice, and it appeared in Top Gun: Maverick. But the bigger debate is now fully open:
Who owns a performer's digital likeness after death?
What does consent look like for future AI use?
Where does tribute end and exploitation begin?
This is just the first film. It won't be the last.
Source: Sky News via Mindstream
Tool of the Week
Google Stitch
Google just launched the design version of "vibe coding." You describe what you want in plain language and Stitch builds a clickable interactive prototype in seconds. No design skills required.
What's new in the latest update:
Voice editing mode: talk to it and watch it change live
Multiple design directions generated at once so you can compare options
Built-in style guide to keep everything consistent
Who is this for? Anyone who needs to mock up a webpage, pitch a visual concept, or explore a rebrand... without hiring a designer for every iteration. It's free in Google Labs right now. If you've been curious about vibe tools, this is one of the most approachable places to start.
Source: The Rundown AI
Quick Hits Worth Your Time
→ Microsoft launched MAI-Image-2, its own text-to-image model, landing immediately at No. 5 on the global AI image leaderboard. Text rendering jumped 115 points, making it genuinely useful for posters, slides, and infographics. Free to try in the Microsoft MAI Playground for US users. (The Rundown AI)
→ Patreon's CEO called out the AI industry at SXSW, saying the "fair use" argument for training on creator content is "bogus." His logic: if it truly counted as fair use, there'd be no reason to cut licensing deals with publishers at all. He's not anti-AI... he just thinks creators deserve a seat at the table. (TechCrunch via Mindstream)
→ Meta is quietly falling behind. Their next flagship model "Avocado" has been delayed until at least May after underperforming internally, and Meta is reportedly considering licensing Google's Gemini as a fix. When a company that spent $100M poaching AI engineers shops for a competitor's model... that's a signal. (Mindstream)
→ Perplexity launched Perplexity Health, connecting health apps and wearables directly to its AI assistant. Instead of generic information, you get answers personalised to your actual data. Early days, but it points to exactly where personal AI is heading. (The Rundown AI)
→ OpenAI and AWS signed a deal to bring ChatGPT into US government classified systems via AWS GovCloud. Anthropic had a head start here through Amazon Bedrock. This is OpenAI catching up in what will be a massive market. (Mindstream)
Prompt of the Week
With AI pricing set to shift toward usage-based billing, now is the perfect time to audit your AI stack and figure out what's actually working for you.
You are a business advisor helping me evaluate my AI tool
investments. Here is my current situation:
Business type: [describe your business in 1-2 sentences]
AI tools I currently pay for: [list them with monthly costs]
Main ways I use AI right now: [list your top 3-5 use cases]
Tasks I still do manually that feel repetitive: [list them]
Biggest time drains in my week: [list them]
Based on this, please:
1. Identify which tools are giving me the most ROI and why
2. Flag any tools that are redundant or underused
3. Suggest 2-3 use cases I am probably missing that would
   save me the most time
4. Give me a priority order for where to focus my AI efforts
   over the next 90 days
Be honest and practical. I want to use AI smarter, not just more.Why this works:Â Most of us adopt AI tools reactively... sign up, use them a bit, keep paying. This prompt treats your AI stack like any other business investment and gives you a personalised audit in plain English. As pricing shifts, the businesses that are intentional about this will have a real edge.
My Take
The dog cancer vaccine story is the one I keep coming back to this week.
Paul Conyngham didn't have a lab or a research team. He had a sick dog and access to the same tools you and I have right now. And he figured something out that researchers are still studying.
That's not a fluke. That's what happens when powerful tools land in the hands of someone who is genuinely motivated to solve a problem.
There's a lot of noise this week about who controls AI and what it's going to cost. But the story that matters most to me is still the dog. Because it's a reminder that the most important question isn't who owns the technology. It's what you decide to do with it.
See you next week.
Jackie
Comments