Michelle Dickinson PhD (@medickinson) header
Michelle Dickinson PhD (@medickinson) profile picture

Kitesurfing engineer passionate about emerging technology, currently focused on physical AI and industrial robotics.

Mountain View, CA
Joined July 2008
Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)Photo by Michelle Dickinson PhD (@medickinson)
I’m so lucky to work at the intersection of AI, robotics, and real people doing real work. I’m interested in technology that’s reliable, practical, and quietly transformative, I’m not interested in hype. I’ll call out nonsense when I see it, and celebrate tools that give people more agency, creativity, and dignity at work. If you’re here for thoughtful optimism and human-centred progress, welcome 🤗
1
20
2,099
Last week I was speaking to a room full of teachers, and the question I was asked most wasn't about curriculum, cheating or which AI tools to use, but instead "What do we actually prepare kids for now?" My answer is that we can't make kids future-proof, nobody knows exactly what jobs will exist in 20 years. However, I think we can help them to become future-ready and I think that starts by focusing less on what machines can do, and more on what we want our children to become. Curiosity - When answers are available in seconds, knowing what questions are worth asking becomes even more important. Empathy - AI can produce a compassionate response, but children learn what it means to care about others through real relationships, shared experiences and understanding different perspectives. Resilience - Getting something wrong, feeling frustrated, picking yourself up and trying again. These are experiences children need to have, not challenges we should automate away. Judgment - AI can sound incredibly confident and still be completely wrong. Learning to question, evaluate and decide for yourself is becoming an essential life skill. Communication - Having a brilliant idea is one thing, but helping someone else understand it, believe in it and want to be part of it is another. Creativity - Not just coming up with ideas, but making something from nothing, experimenting, adapting and discovering what works. Learning through the physical world - Building things, pulling them apart, mixing, pouring, measuring, balancing, breaking and fixing. I was a child who learned best by doing because it allowed me to explore ideas with my hands rather than just try to absorb information. This year I've been working with robots, and while their capabilities continue to improve, I'm more convinced than ever of the importance of giving children opportunities to understand the world by interacting with it in person. The goal of education was never simply to produce children who could answer questions or complete tasks. The goal is to help them become capable, curious, thoughtful humans who can make something meaningful of the world around them. I will never try to predict the world our children will inherit but I do think we can help them to value their curiosity and build their courage so they have the judgment to shape it.
🤖 Made with AI
2
3
19
1,132
Yesterday the Queen of Thailand flew a fighter jet solo, making her the first and only Queen in the world to pilot a fighter jet, having logged around 1,188 hours across several types of aircraft since she began flying in 2010. bbc.com/news/videos/cqm263x4…
8
37
1,078
I've been trying out Hark Pro this week, and honestly, it's been super fun. I use agents all the time at work, but this week I gave Hark a real parenting problem - my daughter starting jazz class (and me knowing nothing about dance). Hark worked out what she needed, found the right sizes, compared prices and paid. Less than 24 hours later, everything arrived and fit perfectly. Commerce is shifting fast and I'd never heard of the store the Hark agent chose. Makes me wonder how agent ready many retail businesses are.
1
7
894
Aged 18-30? New Zealand citizen, permanent resident or UK, US or Irish citizen residing in New Zealand? Applications close October 5th 2027!
Last week to apply! Over the past few weeks, we've shared what it means to become an Inspiring Explorer™. You've discovered the historic expedition bases and their stories you'll step inside. The conservation work you'll contribute to. The skills you'll develop. The lifelong friendships you'll make. The stories you'll share through your outreach. The ‘Explorer Mindset’ you'll grow, and the legacy you'll become part of. Now, it's over to you! We're looking for young people aged 18 to 30 to join the Inspiring Explorers Expedition™ Ross Sea 2027. Alongside young NZ citizens and permanent residents from across Aotearoa New Zealand, places are available for: - One citizen of the United Kingdom currently residing in New Zealand - One citizen of Ireland currently residing in New Zealand - One citizen of the United States of America currently residing in New Zealand There are three types of roles to apply for in the team: - A general role – you have a passion for Antarctica and Antarctic history, and a desire to step outside your comfort zone and embrace new challenges - A heritage professional role – you are studying or working in the fields of conservation, heritage, science, engineering, surveying, or architecture - A videographer role – you have the skills to capture high quality video footage and audio recordings while working in a remote and challenging environment. If you're ready to step outside of your comfort zone, we'd love to hear from you. Apply for the Inspiring Explorers Expedition™ Ross Sea 2027 at inspiringexplorers.co.nz Applications are to New Zealand citizens or permanent residents, and UK, U.S. and Irish citizens residing in New Zealand. Applications close 5 October 2026. The Trust thanks Expedition Partner Heritage Expeditions and donors to the Inspiring Explorers™ Fund for making this opportunity possible.
1
882
I keep having the same conversation with CTO's: "AI means I've basically stopped hiring grads." While it sounds efficient, it might also be how we quietly stop building experts for the future.
8
1
6
1,147
Have you heard the term Physical AI yet? I've just come back from the @AutomateShow with @workr_official and Physical AI was talked about on so many booths. Every person I met selling it described it differently, which tells you a lot about how fast this space is moving. In simple terms, Physical AI is AI that doesn't just live on a screen. It acts in the messy real world, where nothing is ever quite where it's supposed to be. I like to think about it like the person who adapts the recipe on the fly when cooking a meal. A recipe can't see that your pan is too hot, but a cook adjusts as she goes. Physical AI is like the cook, not the recipe. This year there were lots of humanoids on site, and while they get all the headlines and the eye-watering valuations, I'm still a sucker for the quiet little robot arm doing one job beautifully in the corner of a factory. That's what's genuinely working and solving problems for people in factories today. #PhysicalAI #Robotics #FutureOfWork #Automate
6
2
1
9
1,262
The media is obsessed with humanoids right now, but I think that reliable deployment of them at any real scale is still years away. What's actually changing manufacturing floors today isn't a new form factor, it's AI going into the robots that are already there, doing the heavy lifting. At @AutomateShow last week I filmed a handful of robots that show just how different those form factors are, each one solving its own niche problem. A @FANUCAmerica arm hoisting a full car door, a @PathRobotics quadruped welding on legs, a Proxie Cobot pushing a trolley, a @NEURARobotics on wheels working a shelf with two arms and a humanoid handing something to a person. There is no one size fits all, what they all need is the same thing - intelligence. Some of the above robots already have some physical AI integrated into them and this is where I get excited about the future of robotics and physical AI. Intelligence that lets robots pick up a new task in minutes instead of getting pulled off the line and reprogrammed for weeks or sidelined the moment the job changes. Humanoids will have their place eventually, but the robots quietly getting smarter right now are the ones that aren't getting the media attention because they are already doing the work. #PhysicalAI #Automation #Manufacturing #Robotics #automate
1
3
6
561
Michelle Dickinson PhD retweeted
Thanks to everyone who stopped by to chat to @medickinson at the WORKR @AutomateShow stand today. We loved the objects you brought to create a digital asset using our Image-2-Mesh technology. Still time to try it yourself! 📍 @ABBRobotics Booth #1241 📸 Bring an object
1
1
5
240
Michelle Dickinson PhD retweeted
Not all parts look like their perfect CAD file, at @AutomateShow WORKS is showing a different approach. Bring us any object and watch us turn it into a 3D mesh in seconds, from a single image ready for our robot to pick. @ABBRobotics Booth #1241 June 22 to 25
1
3
127
For those who need to know today is National Brisket Day! I’m celebrating - are you?
2
225
I just bought some food. Apparently this is it. America does some great things, this was not one of them 🤷‍♀️
2
4
505
AI has conquered the internet, but what will it actually take to put smart robots in our homes and factories? Here are the takeaways on bridging the gap between foundation models and physical robotics, straight from the Stanford Robotics Center panel featuring @chelseabfinn, @BradPorter_ , @chetan_ and @kaandogrusoz: 1. Generalist "Foundation" Models. Specialists Generalist models are proving more effective at specific domains than specialized models because their diverse training data helps them handle unpredictable edge cases. For example, one robot was trained solely to insert a 2mm pin using its right gripper. During a test, it dropped the pin, but autonomously adapted by picking it up and perfectly inserting it with its left gripper despite having zero training data for that left-hand movement! 2. The 99.9% Reality Check. In massive operations like Amazon, a 95% success rate isn't good enough because handling exceptions requires human intervention that kills efficiency. To actually survive in commercial production, these models must hit a 99.9% reliability standard to build customer trust. 3. The "Hands" Bottleneck. There are currently no truly generalized robots in the world because reliable physical manipulation remains unsolved. Most commercial robots today lack hands, meaning workspaces have to be completely customized to accommodate the machine. Once AI foundation models finally unlock reliable manipulation, the adoption of physical robots will explode just like the adoption of LLMs in software. 4. The Sub-$15/Hour Rule. At the end of the day, robotics is a strict economic equation. With fully-loaded human industrial labor costing between $17- $28 an hour, a robot must be able to perform the work for less than $15 an hour to be a sensible investment. For consumer home robots, the price point has to directly align with the value of the time saved by the user, which requires minimizing parts and simplifying designs to lower manufacturing costs. Great panel and interesting food for thought.
1
2
394
Michelle Dickinson PhD retweeted
We’re building something pretty special at WORKR, and we’re hiring. If you’re someone who cares about building technology that actually gets used in the real world, we’d love to talk. Come and join us as we build the worlds largest robotic workforce: workr.com/careers
3
4
256
Concierge robots are part of the robotic workforce plan and I got to play with this one by @intbotai - while it wasn’t perfect, the technology is still in development and I always have respect for robotics companies who are deploying now knowing they will be learning lots along the way!
1
1
6
844
For those who wanted more Olaf ⛄️ @nvidia #nvidiaGTC
1
466
Look who I found! It’s Olaf from Jensen Huang’s @nvidia GTC keynote talk! nitter.cf/nvidiarobotics/status/… @NVIDIARobotics #nvidiaGTC @nvidiaomniverse @Disney
Olaf made a special appearance on stage with NVIDIA CEO Jensen Huang at #NVIDIAGTC.☃️ NVIDIA and Disney Research are working together to bring this beloved character into the real world using Newton, the open-source physics engine.
1
4
376
Michelle Dickinson PhD retweeted
Olaf made a special appearance on stage with NVIDIA CEO Jensen Huang at #NVIDIAGTC.☃️ NVIDIA and Disney Research are working together to bring this beloved character into the real world using Newton, the open-source physics engine.
134
771
283
4,039
633,854
Michelle Dickinson PhD retweeted
Replying to @ABBRobotics
@ABBRobotics announces RobotStudio HyperReality, built on @nvidiaomniverse libraries, closing the sim-to-real gap. WORKR is proud to be named in this announcement as the company already bringing this type of technology to SME manufacturers. youtu.be/DBlb8VPWaDM?si=AYFM…
3
4
366
Michelle Dickinson PhD retweeted
Be honest, when someone mentions "physical AI" at a trade show, do you nod and move on? We broke it down as simply as we could. What it is, how it works, and what it really means for your team and your manufacturing floor. workr-labs.com/post/what-is-…
1
1
2
230