I'm so grateful and exited to share that my book will be published on June 11th!
It's called "The Skill Code: How to Save Human Ability in an Age of Intelligent Machines"
Available for preorder now, here's a bit of the story:
AI platform finds accident predictors in Waze data, gives drivers real-time feedback, 90% of drivers slow down, accidents down nearly 20%. Seems like benefit might extend to non-Waze users too, but they don’t say...
Kind of a huge deal on the explainable AI front, which is kind of a huge deal for science. Mini-🧵.
We've known for a while that AI finds patterns that humans can't understand. For a scientist trying to build theories of how the world works, this is an obstacle.
Understanding how neural networks reach decisions can be challenging, but is also valuable for uses from image analysis to scientific discovery. A new approach, called StylEx, discovers and visualizes how disentangled attributes affect a classifier →
And so it begins: Germany literally paving the way for designated lanes for robotic vehicles with no drivers onboard. This was the predictable practical solution from the start. Easier to engineer the environment - socially and physically - for a robot than the other way around.
The German Bundestag (German federal parliament) is adopting a new law governing the future of
#autonomous
driving in Germany. Read about the newest legislation:
Uh, Sora from OpenAI about a week ago.
Big, quick debate on whether it was a world model or not.
@fchollet
settled that, imo: not. But verging.
Today, Google swats back with an *actual* generative AI world model?!
This is... unprecedented pace of innovation? Can I say that?
Google presents Genie
Generative Interactive Environments
introduce Genie, the first generative interactive environment trained in an unsupervised manner from unlabelled Internet videos. The model can be prompted to generate an endless variety of action-controllable virtual…
Overjoyed to report my "shitty robot" paper is published
@ASQJournal
!
That'll take some unpacking, but the key is it addresses a critical, everyday organizational problem: how do organizations preserve results given a dynamic portfolio of technologies?
@emollick
If they just embedded a few agents to convince/assist humans to raise funds, run physical experiments, acquire computing hardware, sell product...
It might not be irrational for said humans to follow system-generated prompts.
If an organization ran this way - even in part...
This
@wsj
piece tells an alluring story: Covid means more advanced robotics in warehouses, and big companies win.
@erikbryn
and I tell a different story
@mitsmr
in Sept, based on 1.5 years of nationwide field research. Stay tuned...
@TrungTPhan
What a pleasant surprise, Trung - glad you found this work compelling.
The good news is that a rare few residents got very good at robotic surgery in spite of these barriers. We can learn a lot from their ingenuity.
Here's the original paper:
Had the incredible privilege and challenge of giving a TED talk last November. The idea: we're sacrificing learning in our quest for AI-driven productivity. May it make a difference - learning on the job is more important now than ever!
Brett's business strategy was contingent on not changing his view.
So this belief change - that he will have to justify to funders and potential customers - is a strong signal that AI is making *very* significant and rapid progress in the robotics space.
The timeline split of AI vs Robot Hardware has changed
the last 90 days i’ve witnessed industry leading AI in our lab running on humanoid hardware, and frankly it’s blown me away
i’m watching robots performing complex tasks entirely with neural nets. AI trained tasks that i…
Flabbergasted. Huge honor and life moment to receive such an endorsement from the preeminent group of scholars focused on the implications of advancing technology for organizations! Many thanks to all who supported me in this work - especially the surgeons and residents.
#aom2020
OCIS Best Published Paper announced August, 10 during the Business Meeting. Congratulations to
@mattbeane
"Shadow Learning: Building Robotic Surgical Skill when Approved Means Fail" (ASQ). More info:
Tomorrow's my first class since ChatGPT.
And you know what? I'm excited!
First, it's required. Thanks
@emollick
.
Second, not worried about cheating. Counting on it! I expect better writing and learning.
Third, *I* will use it to improve my questions and feedback.
Here we go!
Thanks for the shoutout, Ethan.
This is my mission. Imo a trillion-dollar problem.
A key cause is we get a short-run productivity boost by empowering experts, but at the expense of novice involvement. The expert-novice collaborative bond is ancient, effective, and invisible.
Training is going to be a key problem in organizations. Triple whammy:
1) Tasks that were given to interns to learn are being done by AI now
2) Less in-person time
3) Pandemic retirements & learning loss
People like
@mattbeane
are thinking about this, but it is a growing issue.
What art can humans and robots make together that no human could make alone?
This is a key question of our age.
Look at this stunning work. A new approach to photography, impossible without sophisticated drones:
@emollick
This is interesting, and I interpret things differently.
First, less than half *said* they thought it was fake after the experiment. This is not proof that they they suspected the shocks were fake *during* the experiment. This could be post-hoc rationalization of trauma. Right?
We have decades of research showing that we build most of our valuable skill working alongside an expert. Yet most the tech we’re developing to help people learn is designed for solo use out of the stream of your everyday work.
Here it is: humanity’s final look at
@NASAWebb
as it heads into deep space to answer our biggest questions. Alone in the vastness of space, Webb will soon begin an approximately two-week process to deploy its antennas, mirrors, and sunshield.
#UnfoldTheUniverse
My team and I see an extreme form of this for warehouse workers - they can't even build meaningful relationships with each other:
"Honestly, I can't [tell you what my coworkers are like] because I don't talk to anybody there. We're trying to go so fast to make our numbers."
Time poverty is a "threat to well-being and economic development that often goes unnoticed." Time poverty lowers life satisfaction & well-being at home and creativity & performance at work.
We are less sensitive to losses of ⏰ than💰 so we don’t notice.
This is my hype up for the weekend.
Over three years in warehouses, big and small, I saw a lot of skill with these things, but very little praise for it. Nice to see everyone get fired up in recognition of this amazing wizardry.
Just discovered the German Forklift Championship. Every single part of this video is amazing: the course challenge, the competition buzzer and the overly-enthusiastic announcers.
A serious problem in biz education right now is the case writeup is dead. ChatGPT can write an A- paper for you!
To help my students, I've used GPT-builder to make a Case Study Coach. It won't give students answers but will tutor and guide them. WIP!
Google now translates poor English —> good English (i.e. grammar and syntax checking) while learning from the best google docs writers. A very, very big deal (threat?) to those who make a living through clear thinking and writing.
Wow! Honored at the shoutout in
@emollick
's new book.
I've been on about this threat - and the shadow learning that overcomes it - since 2014.
By 2018 I had the data to see this was everywhere. That drove my
@TEDTalks
,
@HarvardBiz
, and now book: (June!)
The big education crisis caused by AI is not going to be in schools (there was cheating before AI & we can figure out AI uses that boost learning), but after graduation.
White collar work is secretly based on an apprenticeship system that will break
From my book Co-Intelligence
The Skill Code: How to Save Human Ability in an Age of Intelligent Machines is off to the printer!
Nearly four years ago, I began the messy, ill-advised journey of writing it, and early praise says it delivers.
Pre-order, learn more:
#TheSkillCode
🧬
If you're an ethnographer who studies work involving advanced technology and you'd like to join a team on a nationwide study of AI-enabled robots in warehousing, talk to me. More sites, tech and crazy interesting social dynamics than we can cover!
We don't have good studies that show it yet, but:
Given what we *have* learned about remote work, I think we're in for a doozy of a hit to informal skill development through collaboration.
We take it for granted, but that's the main way we build our most valuable skill.
Learning & Development: the new
#1
organizational function
Organizations can't adapt if their people don't. And even if genAI is "done", we have to learn new skills, soon. That means the CEO's new priority should be L&D, and it needs to step up.
Wow.
For advanced roboticists, pose estimation - predicting the precise location of an object in 3D space - has been a ridiculously vexing problem.
Until now, real-world success has relied on fusing diverse sensor data (e.g., IMUs, cameras, radar, lidar).
Now, just images.
Introducing Category-Level 6D Object Pose Estimation in the Wild.🏞️
We release Wild6D: an in-the-wild object-centric RGBD dataset with 5000+videos over 1700+objects. We perform semi-supervised 6D object pose estimation on it without manual annotations.
Anyone saying "we should teach prompting!!" is now officially on shaky ground.
I've seen three prototypes like this in the last week.
Knowing what a great prompt looks like is still crucial.
But even that may fade: these systems can assess the quality of the output, too.
Holy shit.
@henrycunh
built a UI around `gpt-prompt-engineer`, and it looks fantastic.
You can literally watch it generate prompts and have them battle it out in real-time. 🤯
Prompt engineering is going to transform dramatically.
This is dangerously misleading. We have to get this right. Short 🧵.
Many reputable researchers are saying that AI benefits lower-skilled workers: they can do better, faster!
But they - and their organizations - may often be harmed as a result of their improved productivity.
@emollick
Yes. And training is just the tip of the skill iceberg.
Most valuable skill is developed informally, between experts and novices as they work together. But we take that for granted.
Many will grab for formal training solutions here. That's backwards. New paradigm required.
New publication
@OrganizationSci
with Callen Anthony that's looking quite relevant to our LLM moment. Short 🧵.
tl;dr: How do senior occupation members get up to speed w new tech?
By creating "inverted apprenticeships," where they learn from novices!
Happy Sunday.
It's 2023 now, and this problem remains profoundly unsolved.
And it's coming for your occupation, too: we're sacrificing novices' skill in our technoquest for productivity.
Neither necessary nor desirable.
Research coming soon that shows a brighter path.
The Da Vinci System is a surgical robot that costs $2m. It's been used in >10m procedures and is designed for one surgeon to do the job of many people.
The robot works, but at a cost: the next generation of surgeons aren't getting the reps to learn news skills.
Here's why🧵
THE ROBOTS ARE COMING FOR YOUR JOBS!
We've been saying this for about a hundred years.
And it still gets clicks, chatter and stokes fear.
It's also been recently, decisively proven TRUE, but the size of the effect is TINY. Tiny, folks. Short 🧵.
We're getting a genuine stream of papers genAI * work now.
Here's one on sales. Highest prior performers got the biggest boost - they responded to customers more creatively, drove more sales, mostly because they focused better on serious customers.
Soon after I started my PhD I settled on ASQ as my acid test: if I got published there, I'd be doing work that mattered.
This issue shows why. Superb, consequential research!
I'm a particular fan of the last piece, explaining the resurgence of analog synths in a digital age.
Pop quiz: After a year in a warehouse, your
#AI
enabled robot picks product with 99.7% accuracy. You move to a new customer and it takes another year to hit that target with the same products, bins and work flow. If you're like the 8 vendors in our nationwide study, why is this?
Literally all *empirical* research predicts the economy will adapt to GPT and more jobs will be created than are lost.
But we have daily, rigorous speculation (e.g., the GPT-4 is an early AGI paper today) strongly claiming this time is different.
Stay tuned, and stay frosty.
I've been pretty quiet here for the last six months or so - been focused on getting research published.
Today I heard that a third manuscript (of three) was provisionally accepted. Whew.
This one features a "shitty" surgical robot.
Will share each when actually published!
Random shoutout for a marvelous book on a topic that gets *way* too little attention: maintenance.
We obsess about creating and using things. That includes researchers! We need more studies of upkeep.
@STS_News
@RussellProf
Had the great honor and pleasure of hooding Danielle Bovenburg
@UCSBTMP
(bound for
@YaleSOM
) with Steve Barley. My first doctoral student and his last (we co-chaired her committee), and Steve and I both got our doctoral degrees at the same place:
@MITSloan
!
Watch her work!
Published today: study of AI (yes, actual LLM-style) effects on sales - creativity and outcomes.
Core finding? Workers with most skill and creativity benefit the most. Effect appears nonlinear.
Here's the thing: I really enjoy reviewing papers.
I make sure I'm never overloaded (3 max per quarter), and that I turn away papers where I don't think I can help.
I've learned so much this way! About new phenomena, methods, data, arguments... it's like refreshing my PhD.
Ooof. Consistent with what I’ve found across 31 occupations: we’re using productivity enhancing tech in ways that weaken the expert-novice collaborative bond - where we build most of our valuable skill.
That kneecaps the next generation, right when we need them most.
I feel for people starting in tech now. When I started, I learned directly from experienced designers, product managers, and engineers—pairing and collaborating in the studio/office. It feels like that's been replaced with an abundance of paid courses.
Student:
"The writing process itself really facilitates the development of my ideas. I found myself wasting a lot of time rewriting [GPT]... to get clarity... or nitpicking phrasing that didn’t align with my style. So I [dropped GPT and] just rewrote the analysis from scratch."
Absolutely gobsmacked today to discover (through an incidental conversation with my dad) that my great, great grandfather graduated
@MIT
undergrad (course IV, aka Architecture) in 1893.
Here's his thesis. Thanks for digitizing everything,
@mitlibraries
!
A true honor to be published
@ASQJournal
!
Perhaps more importantly: when it comes to understanding the impact of technology on work, occupations, organizational change and so on, I hope this encourages us to go beyond the "one technology at a time" paradigm.
Discoveries await.
Core finding across my work is we're handling intelligent technologies in a way that makes it harder to learn as we do our jobs.
Billions affected, multi-trillion dollar problem. The time for aggressive, creative investment in skill development is now.
@pmarca
AI certainly won't cause lasting unemployment.
But technological evolutions displace jobs: the faster they take place, the more people are (temporarily) left behind because their skills are outdated for the new economy. Workforce retraining is of the essence.
1/2
@emollick
Great studies of paradigm-shifting technologies (e.g., telephony and gossip) suggest that your warning is basically proof that the tech will be used in exactly the way you warn against. At scale.
I think our job is to hypothesize the particulars and predict the consequences.
Ooo, cool: new work from
@bureaulab
showing how organizational structures (aka bureaucracy) actively shapes the expertise of the people working in it - creating and hiding it, for ex.
Another interesting aspect of organizational friction,
@work_matters
!
@emollick
That seemed like a small gain, so I read the paper, and in human terms, this same shift (from a s*tty manager to an excellent one) translates to a 60% (!) reduction in turnover, with a bias towards "regretted" quits - employees the firm would have preferred to retain! Wow.
Yes! Cc'ing its "inventor":
@AmyCEdmondson
Psych safety is NOT:
-A safety mindset in physical work (get that a lot)
-Where we are "kind" so others don't feel judged (it's compassion WITH accountability, high standards)
Psych safety IS profoundly valuable and hard to create.
I occasionally see misuse of the term “psychological safety.” It isn’t a buzzword, but a research-backed approach to making sure everyone on the team feels comfortable taking risks. At studies at Google it was the key predictor of innovative team success.
Matters: how many jobs we lose because of automation.
Matters more: what new jobs are possible because of automation.
Matters much, MUCH more: How we transform work with our technology (in particular, how much we sacrifice job quality and human capability for productivity).
It's a great pleasure and honor to do my work through such a superb college of engineering.
To have that college recognize me so publicly, then - well, it means a lot. Definitely fuel for the journey.
We don't have to settle for this.
Since 2018, I've published work on ways to build skill even though we're reducing novices' involvement to get value from tech. More forthcoming in the next month or two.
Key point: we can get both productivity and skill. Now's the time to try.
This is unfortunately likely right. Research shows that work from home, though good in other ways, severely reduced mentoring behavior.
And AI is going to mean that managers outsource a lot of the small tasks that used to be sent to young workers to AI instead, lowering learning
I think this is a magnificent study. And there's more to this space than advisor/worker, imo. I think there's at least a proactive v reactive distinction - right now ChatGPT is reactive only. Devin is proactive. Differences could be stark. Or not!
I would love to see more work on AI as advisor to humans, rather than AI just doing work.
This controlled study in Kenya found top small business entrepreneurs got a stunning 15% boost in profits when given an AI mentor, but low performers struggled with mentorship & did worse.
GPT is coming for manufacturing, not just knowledge work. Three-tweet 🧵.
Let's start with design. I asked
@bing
: "Analyze CAD design work. What are the key limitations of CAD approaches, and how might I use GPT-4 to address each of them?"
Real opportunities here:
I have no idea how much attention
@davidautor
's piece on "How AI Could Actually Help Rebuild the Middle Class" is getting, but I know it's not enough.
If there's an economic equivalent to hard sci-fi, this is it. Plausible. Proximate. Urgent.
Robotics has always been gloomy!
Software's *way* faster, affects *many* more people and *far* more profitable.
So. Why persist?
1) Robots are awesome. Seriously. Huge motivator.
2) New use cases are small, but real.
3) No AGI without actuation in a chaotic world.
LLMs have generally sucked the oxygen out of the room, but robotics seems to have been particularly impacted
Been meeting with robotics experts over the last few days, and the mood is gloomy to say the least
Congrats to UCSB Prof and Stanford Digital Fellow
@mattbeane
for winning Best Published Paper from OCIS AOM! Robotic surgery requires new training methods and what he calls "shadow learning".
#Robotics
#medicine
#Education
#aom2020
Can verify for surgery.
In fact, top robotic surgical trainees in my nationwide study did the opposite:
"I watched that video, I probably watched it, I don’t know, 200 times for an hour-long video. A lot, a lot, a lot of hours watching it."
@ASQJournal
Watch & don’t learn: YouTube videos of experts give the illusion of knowledge
“Simply watching one non-instructional demonstration of an expert performing a highly complex skill leads people to become more confident in their ability to perform that skill”
This doesn't look like much.
And that's why it's stunning.
Constructing 3d space from 2d images was an "impossible" problem. This is now *done* at low cost.
Next: constructing cause and effect (i.e. adding time) from a few sequential images.
For good and ill, this is coming.
Agree.
To raise a related concern, I have not seen a prominent economic or sociological prediction (or model) about genAI that takes agents into account.
And it was clear to me from OpenAI's GPT-4 redteaming that they didn't either.
Few saw LLMs coming. And few see agents.
The more I play with early versions of AI agents the more I think that this is going to be a big move forward in the near future. Giving AIs planning and delegation seems to help a lot.
Both Devin & MAGIS agents perform 8x better than base GPT-4 in resolving real Github issues.
Please note: this robot can pick strawberries *mostly* because the entire cultivation infrastructure has been reconfigured to make it far easier to automate picking them.
This is how we automate in general: simplify the task first. Build to suit. Not the other way around.
Well I'll be darned. Quite honored and delighted that my piece is one of
@HarvardBiz
's 10 must reads on AI.
And hey,
@pleonardi1
, you and
@tsedal
are, too (no surprise there). Go,
@UCSBTMP
!
Now available for ACM Members: "HBR's 10 Must Reads on AI (w/bonus article "How to Win w/Machine Learning" by
@professor_ajay
,
@joshgans
, &
@avicgoldfarb
)" audiobook. Understand the future of AI, scale your AI initiatives, use AI to transform your org.
@emollick
Wow! And, dag nabbit: not useful for me yet.
I have gigabytes of operational and personnel related data spanning multiple years from multiple corporations, and would love to dig in this way.
BUT: it's all covered by NDAs.
The day I can upload and process securely? New papers.
Forget the Turing test.
How about passing the bar? And dozens of other language-based certification exams. Tiny sample:
Architecture
Construction management
Critical care nursing
Psychologist
Therapist
EMT
Plumbing
Mechanic
I bet we'll see new, demonstration-based testing.
#OpenAI
's ChatGPT is ready to become a lawyer, it passed a practice bar exam! Scoring 70% (35/50). Guessing randomly would happen < 0.00000001% of the time
@emollick
Second, those who did *not* say they thought the experiment was fake only administered a couple fewer shocks than those who said they thought it was fake.
Pretty strong evidence of the effect this study is known for, yes? Not nearly as strong as originally thought, but still...
If it wasn't clear, it is now: the limit on robot utility is in hardware, integration and maintenance (i.e., the physical robot).
Put positively: This sense-plan-act stack came *way* faster than I expected, and I've been studying ace-startups in this space for three years. Wow.
What happens when we train the largest vision-language model and add in robot experiences?
The result is PaLM-E 🌴🤖, a 562-billion parameter, general-purpose, embodied visual-language generalist - across robotics, vision, and language.
Website:
Even for Ethan, this is supremely well written: entrepreneurs (and, I think, scientists) can do more, better work with these tools.
And his post (and the papers it cites) gives clear tactics.
This convinces me generative AI will boost innovation. Thank goodness: we need it!
For practical purposes, AI is creative.
Three new working papers show that AI-generated ideas are often judged as both more creative and more useful than the ones humans come up with. And they also show you don’t need elaborate prompts to get good ideas.
Join in! Tomorrow, 9am pst I present new research
@erikbryn
's
@DigEconLab
on how workers and automation can complement each other.
“Engineering Skill: How Developing AI-Enabled Robots and Nonprofessional Ability Went Hand in Hand at JointBot”
Register:
We often make the world worse when we predict the effects of technology on work and employment.
Why? We rely on faulty assumptions, yet use these to make (bad) national law and (mis)allocate billions.
Steve Barley and I suggest a new approach in his recent book:
What's wrong with the way we assess the implications of AI for jobs and employment, and what's an alternative? Co-wrote a chapter with Steve Barley on these questions for his forthcoming book. Our core answer: look all the way down the "stack." [1]
We need to adapt faster than ever to handle all the new tech we're creating.
Thankfully, all the new tech we're creating can help us adapt faster than ever.
The catch: Right now, we're handling tech in ways that separate experts from novices. Mostly, that hurts skill.
A real moment to be interviewed for the
@HarvardBiz
podcast about a week before they publish my related article. If you’re interested in the fate of on-the-job learning in an age of intelligent machines, listen away!
This is NOT (thankfully) an article claiming that robots are managing people.
It's the other way around. Yet another piece highlighting a new job I've been calling "fleet supervision": one human overseeing multiple robots, helping when they get stuck:
Wow. Simply, wow.
@UCSBTMP
, I teach "Project Management for Technical Product Development" to our masters students.
@MikellTaylor
joined us today for the 2nd year in a row. Her talk was... FANTASTIC. Her title? "Robots are easy, people are hard."
A pm masterclass in an hour.
CRITICAL new findings on genAI impact in entrepreneurship.
Abstract says it all: business that were already doing well get a 20% boost. Businesses that were struggling saw a 10% *decrement*.
That's right. It hurt them.
And we know a bit about why. More research needed.
🚨 Announcing a new working paper measuring the causal impact of generative AI on business performance in emerging markets!
This is joint work w/
@nickgotis
(who led the project)
@RowanPClarke
@solenedelecourt
+
@orgRem
.
More on our results below 👇 1/
We've been afraid of being quickly and totally replaced by our own tools for a hundred years or more... and it hasn't happened.
That does not mean this time will be the same.
It does mean that anyone who predicts rapid, mass replacement should supply extraordinary evidence.
An excellent opportunity to plug Nardi and Ekbia's Heteromation. Not a fan of neologisms, but the useful idea is that we create tech that automates 10-15% and moves a healthy chunk of work to the customer (e.g., ATMs).
Robot waiters = more of the same.
I cannot stress this enough: this isnt a robot waiter. It does ~15% of a waiter’s job (physically moving items from kitchen to your table) the other 85% (setting items on table, navigating menu, entering orders, etc) is done but you. The “robot” isnt replacing the waiter, YOU are
Replacing a tedious task means the *flow* of the work gets disrupted, so reskilling requirements run surprisingly deep.
Not just for you, but for those who give you work or receive yours.
Learning your new job *and* the new work system is the looming globe-scale problem here.
Asking what jobs AI will replace isn't that helpful.
Instead, recent research thinks about jobs as bundles of tasks. So, it is worth thinking about the tasks we do as part of our job, and experimenting with which ones AI can supplement or replace. Often, it is the tedious tasks.
My career is peppered with failure, restarts and meandering. Very different from the "exemplary performance" increasingly required to get into high-quality undergrad, let alone a PhD program.
#AcademicTwitter
: is there evidence we're paying a price for this? Might be just fine.
When
@doctorow
was a kid, he could screw up in school, but the system had slack -- it gave second, third chances.
Not today, he notes: Kids today face a world where one tiny mistake creates instantly compounding consequences --
He's right, and it's nuts
Folks, we're crossing a rubicon. I gave
@openai
's new chatbot () a rather complex conceptual question from an exam I gave this year in my Master's level Project Management course.
This would have gotten full marks - maybe even a callout in the next class.
I came across this in my dissertation today. It stopped me in my tracks.
Most studies show robotic surgery gets equivalent outcomes to traditional surgery. You read data like this and you wonder about how much skill remains under the hood in the profession...
This often overlooked paper is super relevant as organizations madly try to deploy AI: Tyre & Orlikowski's "Windows of Opportunity" (1994).
tl;dr: organizations will settle on new routines for new technology way too soon, and miss huge upside.
Short 🧵.
This week we've seen a raft of demos in multi-purpose humanoid robots.
Many roboticists roll their eyes at this. It's easy to make a robot do something "at all" vs. "all the time."
Thing is... they're in field trials now. And cost effectiveness is climbing. Rundown in this 🧵.
Want to teach someone why it's easier to change the work environment to suit robots than the other way around? You can do it one move:
@tomscott
's 5-minute tour of an
@Ocado
warehouse. ht:
@DylanKreist
The trillion-dollar problem is that as we deploy automation - and especially intelligent technologies - we weaken collaborative bonds between experts and novices for productivity's sake.
I found this first in surgery in 2019:
But it's not only there:
Experts are scrambling to use genAI as they see fit.
In many orgs this will be forbidden or limited.
Merton (1938!) teaches us some will try anyway.
New paper shows a bit of the how. Cool twist: wolf-in-sheep's-clothing projects. Appear legit. Aren't.
Programming social machines
Generative AI responds to social cues in your instructions. Nobody knows why, or how to handle it, but we do know: our relationship with automation is changed forever.
Having a sinking feeling that your materials and pedagogy won't survive contact with ChatGPT reality?
Pit in your stomach because you think you have to tear it all up? Yeah, me too.
Fear not! Brave, playful pioneers like Ethan are here for us. One day at a time, we'll do this.
We have a new paper on how instructors can use ChatGPT and Bing to apply complex pedagogy to their classes, while making their lives easier, too. There are lots of prompts to try out.
Paper:
Post with an overview of the strategies: