As the great “AI prophet”, our dearly beloved Elon Musk, once said:
“The least scary future I can think of is one where we have at least democratized AI, because if one company or small group of people manages to develop god-like superintelligence, they could take over the world”.
Well, apparently he is right this time, cause he and a few other elite AI hoodlums with big pockets (want a Grok anyone?) are ruling the AI world. But, at least we don’t have to worry that it will have a god-like superintelligence though, because Tim Crook and his cronies from Apple, have told the world that the current reasoning of AI, that we “see” in GPT4o1 is just smoke and mirrors. Just a bunch of parlor tricks.
And I quite agree. (of course Tim C. had to come up with a statement like that, because Apple clearly missed the boat when it comes to generative AI, but that aside, I think he has a point).
Now, that might sound a tad too dramatic, but if you have ever watched ChatGPT or Grok for that matter (too little, too late, I’m sorry, Elon), trying to compose an email that sounds like an alien which is attempting diplomacy, maybe he is onto something.
And this is exactly what I wanted to write about…
“The fascinating world of human-AI collaboration”.
And this didn’t come just from me, because a recent MIT study asked on October 28th:
Are we better together, or should we go back to the days when calculators were all we had to worry about?
If you like this article and you want to support me:
- Comment, or share the article; that will really help spread the word 🙌
- Connect with me on Linkedin 🙏
- Subscribe to TechTonic Shifts to get your daily dose of tech 📰
Buuut let’s not rush to conclusions….
The MIT study, which was published in Nature Human Behaviour, and which I will not link to because LinkedIn will punish everyone that links to the outside world (and you are grown ups anyway who can do their own internet browsing), well this study is about our rather romanticized view of how we are working alongside AI. You know, the one where humans and AI mesh into some horrible cyborg from the Borg.
Well, don’t be scared just yet, my friends, because it isn’t the Borg that we are turning into. The sort of movie, best to compare it with is more of a buddy cop movie where that one partner (think Bruce Willis) constantly tries to steal the show.
A few researchers, called Michelle Vaccaro, Abdullah Almaatouq, and Thomas Malone (no, they are not former drug runners), looked at 106 experiments across three years to see how well humans and AI are actually working together.
And the results are about as mixed as the latest Star Wars “the Acolyte” series reviews (I quite liked the Acolyte btw..).
Like peanut butter and jelly
One thing that keeps popping up in their article is something they call “human-AI synergy”. I personally think that this is the first step in getting us hoomans to integrate with something like Tesla’s Optimus humanoid robot through X’s Neuralink, but hey, who am I to question the intent of the mighty Dr. Evil, right…
You would think that combining human creativity with AI’s reasoning and fast writing abilities would be like umm… peanut butter and jelly.
Wrong.
Vaccaro’s team found out that human-AI teams actually performed worse than just letting one side or the other handle the task. And when they were working together, that led to chaos. But hey, maybe that’s just the price of progress.
MIT’s researchers… they are ever so practical.. They analyzed 370 results from these pre-cyborg experiments.
And their big eureka moment was that Human-AI teams worked okay-ish on creative tasks like generating snappy headlines or designing artsy posters. Because who hasn’t used Leonardo (visuals), Claude (text), Suno (music), Yadayada (whatever) to generate some stuff and post it as it were their own..
Apparently, the AI can handle the work and humans add that little sauce that only we possess, which is …. taste.
But when it comes to decision-making, like detecting deepfakes or diagnosing medical issues, or deciding which client is most profitable, etc. the AI did way better on its own than compared to a collaborative duo.
Maybe humans are just not great at playing second fiddle.
And if you are wondering what the AI world thinks about all of this “groundbreaking” research, our beloved Andrew Ng, face of Coursera’s number one AI training, and oh yeah co-founder of Google Brain (yeaaars ago), has his own alliterative take on this.
Mr. Ng has been quoted saying that: “Hoomans are just too error-prone to rely on alone”.
But don’t mind him.
He is too busy promoting his courses anyway.
Insights and intuition
MIT’s findings show that throwing humans and AI into a team doesn’t necessarily make us any less prone to mistakes.
In fact, it often just doubles the trouble.
Go figure.
The study’s results shows an AI-boosted boost in “creative” tasks. But our brains tend to become, well, dead weight, when things get a tad more analytical.
Why is that, I hear you ask yourself?
One theory is that creative work requires insight and intuition….
Ponder on this for a while..
I’ll wait.
☕🥐
Insight and intuition is just stuff that machines just do not have.
But creative tasks also involve the tedious and repetitive work of course.
I just have to turn my focus inwards, to see myself having to sort through thousands of image edits and generating endless streams of hashtags for the quadrillion postings I have to do on a daily basis.
Now that sort of stuff is where AI shines.
Thomas Malone (the one with the maffia name, and one of the study’s co-authors), he sums it up nicely:
“Let AI handle the background stuff, like data analysis and pattern recognition, while humans do the nuanced thinking.”
So essentially, AI is great at cleaning up the kitchen, but don’t ask it to come up with the recipe. And when you believe Mr Crook and his possy from Apple, it will take more than our lifetime to get towards a super-duper-intelligent-AI anyway.
So basically it’s still human-in-the-loop for all of us, folks!
AI isn’t a magic wand
Oh, and before I forget.. There’s our friends from Stanford University.
They published a paper on AI-human collaboration last year in Proceedings of the National Academy of Sciences (Google it!), and they are happily stating that AI can perform better alone in judgment-heavy tasks because it isn’t bogged down by human biases.
But they also noted a fascinating twist: when the hoomans outperform the AI, they usually perform better together.
It took me a while before I could comprehend what they meant with this.
In other words, humans are good at figuring out when they’re good and they let AI get on the ride to help them assist with something, just like you do with your little kid when you are renovating your house.
But if AI’s in the lead, we will probably just get in its way.
Buuuut, there is also people like Gary Marcus.
He is a very vocal AI critic.
And he said that we are not using AI smartly enough.
He has pointed out that AI is powerful, he doesn’t contend that.l, but us hoomans are often misusing it, and that we are thinking it will magically fix everything.
I know, I’ve been there (read: Gen AI in enterprises – playtime is over | LinkedIn)
Let’s get more quotes in here from the guy. He has a point:
“AI is a tool. And a bad workman blames his tools”.
(For some reason, I keep hearing B.A. Baracus speaking these words aloud in my head…, just add “fool” in there somewhere)
But I think it seems fitting in the context of this post.
If your job involves spotting fake news or picking out abnormalities in medical scans, you are better off letting the AI go solo. And if you’re brainstorming for a new slogan or creating a meme that will go viral? Maybe go ahead and join forces.
I always use the AI to get the creative juices flowing, for say, warming-uppers at ideation sessions, but the real creativity is not in the prompt, you guys, that’s a myth. It’s just impossible to infer something that has never, ever been done before. That’s like creating energy from nothing. Statistically nearly impossible (yet, not improbable). For more info, read: Lobotomized AI* | LinkedIn
Let’s be real, though.
The average office worker probably isn’t diagnosing diseases or creating the next Mona Lisa.
They’re like, ummm… balancing their spreadsheets and writing emails and creating reports all day long (say it very slowly: allll-daaayyy-loonnngggg).
So where does AI fit into all of this
If you ask MIT, organizations that are thinking of using AI to breed with humans, they need to be very cautious about blindly integrating AI into all human tasks.
AI is some expensive shit, after all (think of cash out, energy, CO2). And in jobs where creativity and flexibility matter, it’s AI’s tendency to color inside the lines, and that might not be worth the dough. But when you throw it at routine data-crunching, analysis, prediction, clustering, classification, yadayada just name a few ML classes, the AI is your new best friend – till the end (chucky says hi!).
Another team, this time at the University of Toronto recently published in IEEE Transactions on Neural Networks and Learning Systems that when AI is used for mundane tasks, it frees up humans to do what we do best, ……..
You just fill in the dots ….. , whatever you think it is that YOU do best, ok?
Their research noted that in tests with financial analysts, when the AI handle was doing some data retrieval and cleanup, it allowed for the hoomans to focus more on developing insights and interpretation the results. And that is a fact. That is how I have come to use Claude and ChatGPT as well. Chad helps me with coding (Python) stuff, and converting, and Jean helps me with data analysis and a bit of coding as well (Javascript). The researchers saw a boost in productivity and job satisfaction when people got to use the AI in this way. It’s a win-win, as long as we know when to step back and let AI take the lead.
The MIT study is a buzzkill for anyone who is hoping for the perfect human-AI interbreed/fusion. But the researchers argue this is not throwing the towel into the ring. It is knowing when to play to each side’s strengths.
The idea of a synergized human-AI powerhouse is still quite a dream, but I’m gonna be realistic. Humans are sometimes good at some things, and AI’s good at others, and forcing teamwork isn’t always the answer.
The future, as the researchers see it, is more about strategic partnerships than universal collaboration.
So, next time you proudly announces your new “AI-powered” initiative, maybe do not expect a miracle.
And if you yourself have to work alongside the AI, do not worry about being replaced…. just yet. Just like any working relationship, the secret is knowing when to let your partner shine or when to let them do all of the grutn work while you focus on getting all of the praise.
Because, let’s face it, even the most powerful algorithms can’t beat a well-timed coffee break and some good, old-fashioned human insight.
Signing off – Marco
Well, that’s a wrap for today. Tomorrow, I’ll have a fresh episode of TechTonic Shifts for you. If you enjoy my writing and want to support my work, feel free to buy me a coffee ♨️
Think a friend would enjoy this too? Share the newsletter and let them join the conversation. Google appreciates your likes by making my articles available to more readers.
Leave a Reply