SAIL: The Backlash, On Device AI, A gentle singularity
June 12, 2025
Welcome to Sensemaking, AI, and Learning (SAIL). I focus on higher education and AI.
The conversation around AI is accelerating regarding its transformative effect on life and work. So is the rapidly growing pushback. Critical voices were slow to get started (well, several years ago a researcher was let go at Google for supposed research violations, but possible more to do with shaming bosses - that voice is largely unheard now as if the researcher has completely faded from public dialogue other than in a small camp). There are some recent books published that take a critical stance on AI, but that commentary has a few days in the sun and then fades almost immediately.
I’d love a rich, vigorous, counter conversation to AI hype. What we are starting to see, however, is more of an emotional reaction than a sustained and effective critique. The AI acceleration camp is equally irrational, so we’re in a window of no sane conversations. And yet, the capabilities of AI advances and it’s impact on making individuals more capable proceeds. We are, however, entering the age of AI Rage.
AI and Learning
The Memory Paradox: Why our brains need knowledge in the age of AI This has been making the rounds (even though it was published mid-May). It is the perfect paper if our knowledge needs were as stable as 1970s. Today, the system, the network, is intelligence. As a society, we are capable of exceptional complexity - due not to a different education model, but rather our capacity to store, access, and do novel things with information. No one would look at the world today and say “hey, we need a pedagogical framing of AI that doesn’t include the overall knowledge amplification that technology has provided humanity”. This paper offers exactly that. With that said, it raises important questions of what it means to know today.
It’s the golden age for designers in education. Content is near zero in terms of cost now. Google’s Veo3 (and their Sparkify pilot) will give individuals incredible ability to create content and learning materials. It’s a creators world as barriers to creating have plummeted. A few quick examples here and here and here. Similarly, what is the emotional impact of a short video, AI-generated, that details historic events like the Bhopal disaster? When video is almost as easy to create as writing a paragraph, how does that impact us and our understanding of a topic?
ChatGPT’s vision for campuses. My biggest complaint for universities is the weak response of leaders to meaningfully plan for AI’s impact on learners and on the future of society. This article details ChatGPT’s vision to be a “part of the core infrastructure of higher education”. Great. I also want that. But, it should be in service of the university’s mission in society and vision for learners with agency. Instead, leadership will sign these licenses with NO understanding of the institutional agency they’re handing over to AI companies. This will be online learning and OPMs 2.0. It’s pleasantly enraging that leadership is absent - even the traditional innovation systems in USA higher education are silent and visionless.
We, as a species, are somewhat at an inflection point. Centuries of dehumanizing work driven by the industrial revolution and collapse of social care systems, amplified by the synthetic experiences enabled by the internet and social media, and there are many lonely, isolated, and hurting humans. Which makes reports like this terrifying for our future: “"AI being incredibly sycophantic, and ending up making things worse…What these bots are saying is worsening delusions, and it's causing enormous harm."
AI and Technology
Sam Altman has a new reflection piece: A gentle singularity “2025 has seen the arrival of agents that can do real cognitive work writing computer code will never be the same. 2026 will likely see the arrival of systems that can figure out novel insights. 2027 may see the arrival of robots that can do tasks in the real world.”
With AI boosterism and grifterism having matured over the last few years, the opposition and counter narrative is growing. It appears that the people are mad.
On device AI - from a safety, cost, and environmental angle - is going to be a huge advancement. Google has been delivering with Edge. Apple is not generally seen to be succeeding in AI. But they are also advocating for similar on-device features that will be useful for app developers. Many AI tasks may not be an API call. Instead, developers can use small model tasks on-device. Even an internet connect would be optional.
Taxes, Death, and dropping LLM costs are the only certainties of modern life. OpenAI drops the costs of their o3 reasoning model by 80%.
LLM development is moving at an insane pace (duh). Here’s a six month progress report by Simon Willison.
AI Engineer conference update. This was an informative and deeply technical event last week in San Francisco. Great review of what’s trending in AI adoption. Sadly, very few education (university or edtech) representation.