In this edition: updates on the latest in AI-driven cancer research, ads in ChatGPT, new benchmarking data from various sources, and Anthropic’s Dario Amade on the future of this technology.
This is Q&AI, our blog series aimed at keeping you in the know on updates in the rapidly evolving world of AI. Sometimes, these will be quick updates on new developments in the field. Sometimes, they’ll be tips on tactics, features, or functionality. If you haven’t met me yet, hi: I’m Jen Taylor, CI’s Director of AI Strategy & Implementation, and your (very human) AI BFF.
Q: Where is AI innovation happening beyond business tools?
A: Microsoft released GigaTIME, an open-source model that simulates tumor microenvironments using standard pathology slides.
It’s a powerful reminder that the same AI driving email summaries and slide decks is also accelerating scientific discovery and cancer research. AI’s impact extends well beyond office productivity — and in many cases, that’s where it matters most.
Q: What do we know so far about ads in ChatGPT?
A: There are rumors that OpenAI will be charging a $60 CPM for their ads. The other piece of information they reported is that there will be no conversion tracking—they will only be reporting on clicks and impressions. This is a really high CPM for no down funnel metrics, in my personal opinion.
The reasoning? OpenAI has been really clear that they will not sell consumers’ data. So it sounds like conversion measurement would be more information than OpenAI feels comfortable tracking.
Q: What does the latest benchmarking data show?
A: Based on data from millions of workers, OpenAI’s State of Enterprise AI Report highlights a growing AI inequality gap: organizations that deeply integrate AI are pulling ahead in productivity, and lagging companies are falling further behind.
The key takeaway is not about tool choice. Success comes from organizational readiness, workflow redesign, and cultural adoption — not just buying licenses.
Project Iceberg from MIT moves away from abstract intelligence benchmarks and instead simulates 151 million workers to measure concrete task overlap.
The finding:
- About 12% of total wage value is technically automatable today
- Most of that lives in hidden administrative and finance work, not flashy roles
This feels like a far more tangible way to understand AI’s real-world impact — grounded in tasks, not hype.
Meanwhile, the Anthropic Economic Index confirms a few beliefs we have been holding true on:
- Over half of AI use is interaction, feedback, and learning, not execution.
- Claude is being used for cognitively demanding, white-collar tasks.
- Prompt quality and output quality are almost perfectly correlated.
AI rewards thinking. Training and judgment matter more than shortcuts.
Q: Any other significant bits of research or writing to be aware of?
A: Oh, yes. The CEO of Anthropic, Dario Amade, published a paper entitled “The Adolescence of Technology: Confronting and Overcoming the Risks of Powerful AI.”
It really talks about the two paths we have with AI: one where we regulate it, we take responsibility for it, we’re thoughtful, we’re cautious. Or…what can happen if we don’t.
He puts forth a really pragmatic framework where he defines what powerful AI is, and he asks us to take the risks seriously and to address them through careful alignment research, transparency, guardrails, and narrowly targeted regulation. His call to action is for technologists, policymakers, companies, and citizens to urgently tell the truth about the stakes, to invest political and moral capital and sensible safeguards now, and to act with courage and restraint.
Dario Amade also appeared alongside Demis Hassabis (co-founder and CEO of Google DeepMind) at the World Economic Forum in Davos. While they differed on their timelines to AGI (artificial general intelligence) they both agree that we are on a near-term path to more intelligent/capability AI models than what we have today. The hard part is not the technology, it’s the labor disruption, institutional lag, and adapting fast enough to keep pace.
I really try not to feed into ‘AI hype’, but when the heads of major AI houses are telling us the pace of change will be faster than anticipated – I think we should listen.
It’s been busy, so you’ll hear from me again soon!
Jen
Have a question for a future edition? Submit it here!