Alden Globe

How do you measure if your AI project has a return? I wrote a 4 + 1 Framework on AI Factory Economics that gives CIOs the language to measure the ROI. https://lnkd.in/g2tWycNt

1

I think there is a shift in the narrative around AI agents. Perhaps they are not going to be autonomous and replace humans. There is increased chatter that leans toward collaboration, rather than automation or even augmentation. LLMs are great at interpreting human intent and producing outcomes that make sense with us. But they may not be so good at being left alone. And that seems to be the way things are going. The problem is no longer about engineering autonomous agents. It is about designing collaborative systems. Collaboration not with tools, but with interpretive AI systems (as I discussed in an earlier post). I wrote a blog post about this shift. I think we are moving away from talking about autonomy, and starting to talk about aligning human intent in interaction. Glad midsommar! :)

33

2 Comments

The big thing here for me was "What does Paul Roetzer think?" Bringing some clarity to another bold claim about the disappearance of an entire discipline due to AI — with another random timeline.

3

𝐋𝐞𝐚𝐝𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐒𝐭𝐫𝐚𝐭𝐞𝐠𝐢𝐜 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐀𝐈 Strategic planning for CXOs is entering a new phase — one where Agentic AI moves Strategic Planning from analysis to active decision support and validation Strategic Agentic AI provides Execs ability to - • Explore and validate strategic scenarios through guided conversational interfaces • Drill into connected source systems for contextual intelligence • Extend existing AI investments into strategic analysis My recent demo on this approach using customer-specific custom-AI agents was highlighted by our product leadership Matt Bradley in an internal blog. Sharing the redacted post here. #AI #EnterpriseAI #ExecutiveAI #StrategicPlanning

9

Words are seldom broken on book covers. Some might call it design heresy. But James Goggin and I did this with "Intelligence" on the front cover of “What Is Intelligence?” (out via The MIT Press and Antikythera on September 23). In part 2 of our conversation, we explain the thinking behind this unconventional choice. The typeface mirrors the topics in the book, which traces how computation has driven major evolutionary transitions across billions of years on Earth: from eukaryotes and multicellular life to the rise of urban civilization and AI. Communicating ideas of this scale requires a font as big as the subject itself.

26

7 Comments

AI transformation only works when it’s driven from the middle, not the top. In our conversation with Barbara Wittmann, she explains why real progress happens when you empower the quiet executors who break silos, rethink assumptions, and spark continuous innovation. Mindset before machines, because AI succeeds when people do. https://lnkd.in/grhNrewp Microsoft Innovation Podcast

5

2 Comments

When most people talk about AI today, they think of ChatGPT. But the story of artificial intelligence stretches back centuries — from ancient myths about automatons to 20th-century research exploring machines that could reason or learn. Even then, humans were trying to offload both physical and cognitive tasks to tools we create. One lesson from that history is that AI has always been about two complementary approaches: Systems that learn from data and patterns Systems that reason explicitly using rules and knowledge Understanding both approaches—and how they can work together—is critical for anyone trying to build or apply AI today. Real intelligence isn’t just about recognizing patterns; it’s about grasping structure, adapting to change, and following rules when necessary. Take games as an example. Humans can easily adapt when rules change mid-play. Most AI systems today still struggle with that kind of flexibility. Designing AI that can both learn and reason opens up possibilities for more robust, reliable, and interpretable systems across industries. Another insight from history and current practice: calculated experimentation matters more than perfection. Early adopters of new technologies often learn the most — even if their first prototypes fail. For organizations exploring AI, small-scale experimentation combined with structured learning can pay huge dividends. Finally, technology is most powerful when it enhances human capabilities. AI doesn’t need to reason exactly like humans. Instead, it can help us reason better, make smarter decisions, and tackle problems that would be too complex to handle alone. For anyone working with AI — whether in research, business, or education — the takeaway is clear: study the past, embrace both learning and reasoning, experiment thoughtfully, and use AI to amplify human potential. A special thanks to Alec Crawford for a conversation that helped crystallize these reflections. #ArtificialIntelligence #AIResearch #KnowledgeGraphs #DeepLearning  #SymbolicAI #Innovation #HumanReasoning #NSF #AIInsights https://lnkd.in/gXpFPYuV

28

3 Comments

Information overload makes it hard to feel confident in a final report. Researcher now uses multi-model Critique (generator + expert reviewer) and Council (side-by-side reports + judge summary). Read more and try it now in the Frontier program: https://msft.it/6049Qfc2L #MicrosoftEmployee

6

What does AI actually look like today—not just in theory, but in practice? According to The Washington Post, Matt Calkins stands out as one of the most credible technical voices in the public arena—bringing decades of hands-on AI experience and a grounded view of where the technology truly is. Read more here.

1

So proud to see this blog from Paula H. Whitfield, MBA — one of our leaders at ImagineCRM. She names something we hear often: teams want innovation, but what they really need is space — to try new things, explore Salesforce AI, and make meaningful improvements. If you’re thinking about how to scale impact and free up your team’s time, this is a great read. ImagineCRM Salesforce Salesforce Partners

6

Researchers introduce interpretable constructs for human organizational preferences in object arrangement, guiding AI planning with spatial practicality, habitual convenience, semantic coherence, and commonsense appropriaten… Photo by Pixabay on Pexels https://lnkd.in/g-tijJbw

1

AI continues to improve exponentially while enterprise scale adoption has been stalled. I write and speak often about data fragmentation, data governance gaps, and other data issues being significant blockers of successful AI deployments. But the leadership needs and challenges deserve significant attention too. I found this HBR article to be relevant and timely.

6

I recently found some MIT research that leads with the headline, nearly 95% of AI-for-business efforts fail to deliver real impact. Not because models aren't good, but because the systems aren't built for scale. Two recent articles (links below) caught my attention because they surface some real blockers that leaders and teams must address: 1. AI pilots fail when systems can't retain context, learn over time, or integrate into real workflows. 2. Bounded context windows and fragile memory architectures lead to inconsistency, rework and stalled adoption. 3. Great demos don't often evolve into production grade systems. So what works? Durable architecture, persistent memory, and business aligned learning loops. If we want AI to move beyond the pilot stage and drive measurable business outcomes, we have to design systems that remember important things, evolve, and compound value over time. How are you all thinking about this as you push your boundaries?  https://lnkd.in/ezQ_Zj5f https://lnkd.in/ekJSqNDN

34

5 Comments

Next week, we’re hosting a Global Accessibility Awareness Day (GAAD) event at Tria focused on how accessibility is evolving across design, development, and AI-enabled experiences. The sessions cover everything from designing accessible experiences from the start to using AI to shift left on Section 508 and automate remediation. Join for a specific topic or stay for the full afternoon. #GAAD2026 #accessibility #uxdesign

25

I have been thinking a lot about the friction between human intent and software execution. This week at State Change, we explored voice-driven intent flooding—using high-volume transcription to give AI the context it needs to find the signal in our noise. It is a more authentic way to move fast. We also discussed using AI as a mirror to identify contradictions in a client's vision before we build. When we move away from mass-market SaaS toward a customer of one operating system, software becomes a bespoke environment that actually fits the individual. Full notes here: https://lnkd.in/e--3rpaJ

2

Axonis applies this idea with AI to make decisions data-driven, traceable, and continuously improving. Axonis Sheth Sanket #axonis #artificialintelligence #ai #transparency #datadriven

4

We launched the AiPEX Center less than a month ago, and the waitlist is already growing. The interest from the Higher Education community has been tremendous. Institutions are eager to see what AI can do to their finance, HR, procurement, grants, student services, and campus operations, in a fully immersive AI experience setup! That is exactly why we created the AiPEX center and the AiPEX University: to give colleges and universities a hands-on way to explore practical AI use cases, ask the right questions, and start shaping a roadmap for responsible adoption. Excited by the response so far, and even more excited for what is ahead.

14