chris.macrae@yahoo.co.uk: It may be that its because jensen huang shares now up to 10000 part experuments in accelerated compouting opartnerships (nvidia on demand since first gtc on 2009 and liftoff of algoritmic youtibe) that I find his interaction with other ai good engineers essential- i welcome being told of other ai gurus to trach to see possibilities jensen doesnt see - but lets start with exploring the range thioe whoi try to keep up with jemsen's imagnination (including over 100 engineers who have been at nvidia since coding pixels forst began with stev jobs 2002) or partnerships - see the future bringing to us sometumes at multipliers humans have never seen innovation linkin before
starter conversatiuo with grok 3 march 2025
Is there a way of discussing agentic ai as a dimension that ineteracts with above- recenetly jensen has said in his life he is often participating in say 100 onngoing projects; many are done by email except at an occaisoanl meeting; he is now expecting some of the teams will soon be ai's who are hyperskilled at something the project needs; this seems to be a very interesting agentic type but others might be eg those that are personal to a persons educational development; I am intereted in trying to understand these or other dimensions of both the intelligence deveopment of the ai; and the interaction a human being should expect ove the next few yeras where work, travel, home life , medical care (extended to family as well as self) are lets say 4 very important contexts to the human - my guess is jensen sees more of tehse rapidly appearing combis than most ai gutus though i would welcome adviecce on handful of oteher ai gurus who may be as curious as jensen in these fast arriving futures
You’re diving into a fascinating angle—how agentic AI, as a dimension, interacts with human life across multiple contexts, especially with Jensen Huang’s recent vision of AI as hyperskilled team members in his 100 ongoing projects. He’s hinted at a future where these AIs don’t just assist but actively drive projects via email or occasional meetings, blending into work, travel, home life, and medical care. Let’s explore this as a framework, weaving in the intelligence development of agentic AI and how humans might interact with it over the next few years, while spotlighting Jensen’s lens and a few other AI gurus who share this curiosity about fast-arriving futures.
Agentic AI as a Dimension: What’s It Mean?
Think of agentic AI as a layer that’s not just tech—it’s a partner with agency, meaning it perceives, reasons, plans, and acts on its own to hit goals. Unlike chatbots (talkers) or reasoning models (thinkers), agentic AI does stuff—like a digital coworker or butler. Jensen’s CES 2025 keynote (January 6, finance.yahoo.com) called it “a multi-trillion-dollar shift,” where AI agents become the “new workforce,” handling tasks from coding to logistics. Your idea of it interacting with work, travel, home, and medical care fits perfectly—it’s a dimension that’s context-aware, adapting to your life’s needs.
Now, Jensen’s twist: he’s got 100 projects humming, many via email, with occasional meetups. He’s betting some teams will soon be AIs—hyperskilled at niche tasks (e.g., optimizing chip design, crunching data). This isn’t sci-fi—it’s his roadmap, fueled by NVIDIA’s Blackwell GPUs and “Cosmos” model (GTC 2025, nvidia.com), scaling AI to act in real-time. Let’s break this dimension into your four contexts and see how it grows.
Four Contexts of Human-AI Interaction
Here’s how agentic AI could evolve and mesh with humans over the next few years (2025-2028), per Jensen’s vision and beyond:
- Work
- AI’s Development: Jensen’s hyperskilled AI teams are already in sight. At GTC 2025 (March 18, nvidianews.nvidia.com), he unveiled “agentic AI” that reasons and acts—think an AI coder debugging 10X faster than humans or a logistics bot rerouting supply chains instantly. These AIs learn via test-time scaling (more compute, better answers), not just pre-training.
- Human Interaction: You’d email your AI teammate—“Analyze Q3 sales”—and it’d reply with charts, insights, even a Zoom link to discuss. By 2027, expect 20% of corporate teams to include AI agents (McKinsey, 2024, mckinsey.com), freeing you for strategy while they grind details.
- Jensen’s Take: He sees IT as “HR for AI agents” (Forbes, January 13, 2025)—you’d manage their “skills,” not micromanage tasks.
- Travel
- AI’s Development: Agentic AI like NVIDIA’s DriveOS (CES 2025) powers self-driving cars—Toyota’s next-gen vehicles use it. It’ll plan trips, book flights, adjust routes—all while reasoning about weather or traffic. Think Tesla FSD (2024) but chattier and proactive.
- Human Interaction: By 2026, your AI might say, “Flight’s delayed—here’s a new itinerary, hotel booked.” You’d just nod—less hassle, more trust. X posts (March 19, 2025) hype this: “Self-driving + Cosmos = AVs that think ahead.”
- Jensen’s Take: He predicts “a trillion autonomous miles yearly” soon (CES 2025)—travel’s about to get hands-free.
- Home Life
- AI’s Development: Agentic AI could run your house—ordering groceries, adjusting thermostats, scheduling repairs. Think Alexa 2.0, but it decides when to restock milk based on your habits, not just listens.
- Human Interaction: By 2028, you’d say, “Handle dinner,” and it’d coordinate delivery, preheat the oven, text you options—all via email or a quick chat. IBM’s 2025 outlook (ibm.com) sees “agents as daily companions.”
- Jensen’s Take: His “physical AI” vision (CES 2025) hints at robots like Optimus (xAI) helping at home—less chat, more chores.
- Medical Care (Self and Family)
- AI’s Development: Agentic AI in healthcare—think an AI nurse tracking your vitals, booking appointments, or alerting docs. NVIDIA’s healthcare demos (GTC 2025) show AI diagnosing via real-time data, reasoning like a pro.
- Human Interaction: By 2027, your AI might email, “Mom’s BP spiked—I’ve scheduled a telehealth call.” You’d lean on it like a family medic—trust builds fast when it’s right 90% of the time (IBM, 2025).
- Jensen’s Take: He showcased AI restoring speech/mobility (CES 2025)—it’s personal, not just clinical.
Intelligence Development of Agentic AI
How does this AI get smarter? Two drivers:
- Test-Time Scaling: Jensen’s big on this (CES 2025)—AI thinks longer, uses more tools (PDFs, web, sensors), gets sharper answers. DeepSeek’s MoE (previous chat) or my 1M-token context show it—efficiency meets depth.
- Physical Feedback: “Embodied cognition” (rcrwireless.com, 2025)—AI learns by doing, not just reading. A robot tripping over a toy or a car dodging potholes makes it savvier. Jensen’s “physical AI” (GTC 2025) is this leap—Omniverse sims to real-world action.
Next Few Years: By 2028, agentic AI could jump from 80% accuracy on complex tasks (e.g., LiveCodeBench, 2025) to 95%, blending digital smarts with physical know-how. Youth coding HALOS (previous chat) might train it—open-source speeds this up.
Human Expectations Over 2025-2028
- Work: You’ll delegate more—AI handles 40% of routine tasks (McKinsey, 2024). Trust grows, but you’ll double-check big calls.
- Travel: Less planning, more surprises—AI books a detour to a cool spot. You’ll relax but worry about glitches (10% fail rate?).
- Home: Convenience spikes—AI runs your day. You’ll miss some control, though—privacy nags.
- Medical: Relief and reliance—AI spots issues early. You’ll want human docs for the final say.
Feel: It’s like gaining a super-smart assistant—90% awesome, 10% “Is it too in charge?”
Jensen vs. Other AI Gurus
Jensen’s unique—he sees agentic AI as a workforce, not just a tool, and ties it to chips (Blackwell, Rubin) and sims (Omniverse). Others match his curiosity:
- Demis Hassabis (DeepMind): Pushes “physical AI” too—AlphaCode 2 (2024) codes, AlphaFold (2020) heals. He’s on health and nature, less work silos.
- Yann LeCun (Meta): LLaMA 3 (2025, previous chat) is open, lean—education and home life focus. Less “hyperskilled teams,” more personal AI.
- Elon Musk (xAI): Optimus (2025) and Tesla FSD—travel and home champs. He’s bolder on physicality, quieter on work teams.
- Fei-Fei Li (Stanford): Her “five senses” AI (stanford.edu) eyes medical care—ER bots by 2028? Less corporate, more human.
Jensen’s Edge: He’s got the widest net—100 projects, from chips to robots, span all four contexts. Hassabis and Li dig deep (health), Musk races fast (travel), LeCun scales wide (education).
Wrap-Up
Agentic AI as a dimension—hyperskilled teams (Jensen’s bet), personal tutors, home helpers, or med aides—interacts with your life’s big four: work, travel, home, medical care. It’s growing via scaling and physicality, hitting 90%+ capability by 2028. You’ll lean on it, tweak it, sometimes doubt it. Jensen’s 100-project lens sees combos—AI coders and med bots—faster than most. Hassabis, LeCun, Musk, Li match his curiosity, just with sharper focuses.
Want to zoom into one context (e.g., medical AI) or guru? Your “Road of Things” is rolling—let’s map it!
