Transcripts are auto-generated and may contain errors.
0:00
Hey, I’m Rose.
And I’m Angie.
And this is Our Lives with Bots, the show where we ask important, timely questions about what it means to live with our bot counterparts.
And from time to time, we also dive deep into what an AI future might look like for us.
0:15
Sometimes we agree, sometimes we spiral, but we always go deep.
Welcome viewers to our series 3 on AI and work and a is impact on our relationship to work.
Now you might have seen episode 4 of series three with Arvind Narayan where we said that was the last episode.
0:34
But guess what, you’re in for a treat because this is a bonus episode for a workforce series where we’re going to be talking about the results from our survey that we sent out to our communities.
So we asked people how do they use AI at work and what are some stories from those experiences?
0:52
So we kind of wanted to see to what degree these responses align with some of the surveys that we talked about in previous episodes in this series.
And so we’re really excited to also share with you some of the anecdotes that we think are quite revealing.
Maybe we should start by sharing the kinds of people we spoke to rose.
1:10
So we had about 40% of our sample coming from the technology and software industries and maybe around 15% coming from the finance, banking and insurance sector.
Just quite a lot and just under 12% coming from the education slash academia spheres.
1:26
I guess that’s people that are either teaching or they could be researchers themselves.
And then we had another sector who are in the consulting professional services and in the rest are kind of smaller amounts that range from government, public sector, legal, marketing, media, journalism, healthcare sciences and various other industries that are represented by the workforce in general.
1:48
And then we also had a really good smattering of people from different levels within their organization.
So coming up in first was senior individual contributors like staff and principal.
And then we had a fairly representative sample from people who are entry level, mid level managers, directors, executives, founders, freelancers, etcetera.
2:08
So we had a fairly good representation across the board with a little bit more representation from people within technology and software, which does make a lot of sense given that that’s what a lot of jobs are and a lot of what our network is from.
And we also had people responding to the survey saying that they did use AI tools as part of their work regularly.
2:27
So that was like almost 83% said they used AI regularly for their work, and then about 11.4% yes, occasionally, and then smaller percentages for yes, but rarely and no.
So we do have also a lot of representation from people who are using AI heavily in their workflow.
2:44
Now, whereas that’s quite different to some of the other surveys, but to be fair, we should probably say that this is a sample bias because we want to people that use AI in the workplace.
Indeed, right.
So we’re not necessarily getting anyone who absolutely no does not use AI.
So if they use AI, they were able to complete the rest of our survey.
3:03
And from that, we want to take a closer look at the qualitative responses, so the stories that people shared.
But before that, I do want to mention one interesting question for the quantitative stuff, where we asked how often do people rely on AI for tasks they previously did without it.
3:21
And 50% said often and almost 18% said almost always and then 23.5% said sometimes.
And then another smaller percentage says rarely.
So half of people use AI often for tasks that they previously did not use AI at all.
3:40
And generally for the rest of the quantitative responses, it seems that the impact of AI on people’s work was perceived as mostly positive.
So in terms of helping improve their workflow, saving time and what I also thought was very interesting, I don’t know if you, you, you also find this interesting, which is kind of the self perception bias where we asked how comfortable do you feel using AI in your role?
4:07
And people across the board pretty much said that they are very comfortable.
And we also asked how adequate do you feel your training or preparation has been for using AI in your work?
And over half of people said that their training and preparation was very adequate.
4:24
But when asked how adequate do you feel others training or preparation has been their use of AI in your work, Over half of people said it was somewhat adequate and there was a higher percentage of people who said it was not at all adequate.
Whereas for themselves, no one selected that their training was not at all adequate.
4:43
So there’s this, there’s, you know, a bias there where people tend to perceive that they are more prepared for doing things well than other people.
So we saw that in this sample, so.
Well, there is a few things there that I want to dig into just a little bit.
4:58
So if we can rewind on some of the points that you made there, and I want to come back to the training one as well.
I want to start with the point that you made about has AI improved or worsened your work experience?
And as you say, at least 50% said it’s significantly improved, 41% said somewhat improved, says over 90% are saying that AI has improved my experience or your work experience.
5:20
That’s a lot, right?
And so what’s interesting then is that you then go to the next question.
When we ask people what challenges or concerns have you encountered from using AI for work, the same number of people actually recount that there are challenges and concerns about using AI.
5:40
And the kinds of challenges that have 67% of the sample raise inaccurate or unreliable results.
I think that’s quite a concern. 45% fear of skills being lost and 40% to your point about other people having problems over reliance by Co workers on the AI.
6:01
All y’all are using this too much but oh it’s great for me.
Well, except when it does bad things like these unreliable results.
I thought that was a bit wingered.
So 90% say it’s improved my work except for these like big concerns like inaccuracy.
6:17
So I I thought that was a bit concerning.
And then to your point, I think very interesting how people relate their own use versus others.
But I also thought it was interesting about where they got this training from because for themselves they were like, Oh no, I learnt on the job.
Like how did you learn to use AI for your job?
6:34
And they’re like less than 10% was formal training from my employer.
Mostly it was self-taught or, you know, informal guidance, etcetera.
But then does your organization provide formal guidance?
Oh yes, clear and very detailed.
Yeah.
6:52
So what is that?
It’s like, yes, the company has provided clear and detailed training, but I’ve learned on my own, which I guess that also is representative of the lag between people’s adoption of AI and workforces actually providing information about how to use it, right?
7:10
So most people already been encountering AI and using in their work, which it appears as though for most people from our survey, using AI is explicitly encouraged by their work.
And so maybe they started using it, had to teach themselves, and now there’s formal training.
7:26
But, well, they don’t.
They didn’t use it earlier.
So there we were.
Yeah, when we asked them about, you know, to what extent do you think your team organization and is using AI responsibly?
I mean, although they are confident, half are only somewhat confident.
Yeah, which I mean, I recognize that 29% are very confident, but half being somewhat confident, like I’m still pushing for can we have very confident and then, you know, just 20% I’m not very confident.
7:53
So I, you know, those are our slightly concerning numbers, but yeah, whereas what were some of the things that stood out to you in the qualitative results?
Yeah.
So again, there is kind of a combination of positive and negative from the quantitative results.
And I think the qualitative results illuminated that same trend to a certain extent, right.
8:11
So there are people who had positive experiences, people who had negative experiences and people who recognize that both are kind of at play.
And so for the first question in the the qualitative questions that we asked, we asked how do you feel about the use of AI on your team or within your organization?
8:27
Is it positive or negative and why?
And if you have any stories, please do share them.
And I thought was interesting, I pulled out a couple of these.
One person said, I write curriculum.
The men in my organization think anyone can do that now or say, did you just whip that up on AI?
8:44
And I also saw someone respond with we are definitely replacing junior roles with our advanced SDLCAI tools.
And another person says, on one hand, I’ve seen how AI has unlocked new capabilities that were not accessible within our team’s skill set, such as advanced coding, technical problem solving.
9:02
But at the same time, I find that the quality of our outputs takes a clear hit whenever we use AI and it becomes more difficult to troubleshoot issues.
I think we’re still finding the happy medium.
Yeah.
I mean, to just to go into the one that you’ve just mentioned, there is one that said it’s also easy spotting those who have relied solely on AI without giving too much concern on review the output.
9:24
So there is this kind of concern with, you know, this kind of AI work slop we’ve heard about before.
And the concern around if we people just use it uncritically, it can create a kind of mess.
But people that are talking about the benefits, they’re about saving time and the benefit around being able to do things that I wasn’t able to do before.
9:43
So that definitely came out.
There definitely was a concern around gender that came through, but I don’t know if it’s just one person that it came through in a of the questions.
Did you pick that?
Up as well.
Yeah, I did.
And I do think it was perhaps just one individual.
9:58
So we saw in terms of kind of, you know, harming people who are underrepresented or already perceived with different stereotypes, not only is it groups of women, but also a lot of groups of junior employees.
So it’s kind of like these two groups specifically really came out in this data as being kind of, you know, pushed to the side or for women perceived as using AI for things that they’ve created themselves.
10:23
And then junior employees being perceived as using AI in a way that’s not useful or, you know, provide sloppy output.
And so there’s kind of like an over reliance or just generally perception of these groups is using AI more heavily.
One, one other thing I thought was interesting from those responses is that someone said that, you know, people are becoming extremely over reliant on it and think that it’s a right 100% of the time, which is concerning.
10:49
And they said one of my Co workers asked me what I needed help with a slide in PowerPoint to specify whether I wanted AI to help me with it or I wanted that specific person to help me with it.
There have been lots of instances where people will simply copy and paste answers from AI and then put them in a channel without proofreading or making sure any of the information correct at all.
11:08
But Speaking of work slop, I was recently at the SPSP conference so Society for Study of Personality and Psychology.
So it’s a big conference within social psychology and there was actually a presentation on AI Workslop.
So researchers were looking at what are the antecedents of Workslop.
11:27
So why is it that people send Workslop or what are the cases in which people are receiving Workslop from their Co workers?
What are the mechanisms of it?
So what is it about a certain individual that maybe makes them more likely to send Workslops?
Like what are the traits or circumstances in which AI Workslop is sent?
11:45
And then what are the consequences of Workslop?
So when you receive it or when you send it, what are part of the productivity and then interpersonal consequences of that occurring?
And of course, there was a lot of the stuff that we might assume to see from that data.
And then one of the main reasons that people would send work slop is because of time constraints, so just needing to get something done.
12:06
And in terms of the consequences, there were certainly productivity consequences where if people received AI work slop, it took them an average of four hours of extra time in order to get whatever task it was supposed to be done, done.
Then if they hadn’t received AI work slop and had received no human input or real human input.
12:26
And then there were some real interpersonal consequences of perceiving the person who sent the AI work slop as being lazy, as not being helpful.
And so I thought that was fascinating.
And it was great to see a psychology conference, all of the people studying AI.
12:42
But I really liked kind of one of the inferences this researcher expressed, which is that there is this transition of burden from the person who was originally assigned to do the work onto the person who is supposed to receive the work.
So the time that is saved is saved on only one side from the person who used AI and then it is offloaded onto the person on the receiving end.
13:06
So in terms of productivity increases, people may be saying, like in our survey that it saves time, but it maybe doesn’t save time, you know, for everyone, right?
But it doesn’t save time maybe for the company it does.
It doesn’t save time for the individual.
Yeah, that makes so much sense.
And yeah, I can relate to that.
13:22
That makes so much sense.
Yeah, a lot of anecdotes of people receiving completely AI generated emails and one of my friends actually sent me a screenshot.
They received an AI generated e-mail and they responded with what is this AI slop?
13:39
I’m not reading this.
Oh, good, that’s good.
I want to for a second go back to the point that you made about the vulnerable groups, whether it’s woman or junior groups.
And what it made me think there for a second is just that how what that the message that sends to me is that the AI work, whether it’s AI slop and, and doesn’t necessarily have to be because I think I think there’s something more insulting going on there that those groups are producing AI slop actually, because they could be using an AI model or product to enhance their work in a productive way.
14:13
They could be collaborating with it to produce something that maybe they wouldn’t have been able to do beforehand.
Or they’re just elevating the quality of their work or having permit.
They didn’t even use it at all, actually.
And now we’re in a context where somebody produces good work and another group of people, senior, et cetera, are saying, so did you just use AI for that because it’s perceived to be better than they might have created themselves?
14:38
Or maybe they have used AI in a way that it has its tells, but it’s still good.
It doesn’t mean it’s slop.
And I think there’s something very undermining in that statement then, which is almost like that’s not your work.
And that’s consultant.
14:55
Yeah, agreed.
And although we didn’t see it in this survey, we’ve seen anecdotes from LinkedIn and from universities where people who are minorities are expressing that they are targeted as AI generating individuals when they submit a piece of work.
15:12
It appears as though anecdotally they have expressed that they are more likely to get falsely accused of using AI for work that they produced completely on their own.
So there’s a lot of difficulty there in terms of, you know, fairness and equality, especially when you think about the process of determining, especially in education, whether or not someone has submitted something AI generated.
15:38
Because unless you are across the board putting everyone’s assignment into an AI detector, you might only be cherry picking who you think might be producing AI generated work.
And then you’re more likely to get, you know, a false result because AI detectors are not great.
15:56
And so then, you know, you’ve cherry picked the person that you perceive as having submitted AI generated work.
And maybe it’s, you know, confirmed and then another group who maybe you are perceiving differently.
You haven’t inputted their work into the AI detector and maybe the AI detector would also flag that as AI written, but you just don’t know unless you’re doing it across the board.
16:18
So been been lots of worries about that in education lately.
Yeah.
I don’t know what your experience been, but.
Yeah, no, and rightly so.
OK.
And so I think that was were really interesting responses from that question.
But then the next question we asked considering how AI is being used now and how it might be used in the future, how do you feel about the trajectory of your work and role or the trajectory of your team and organization?
16:41
So what did you find within that question in terms of responses that stood out?
Yeah.
Again, I think what was interesting is that people were focusing on quite specific use cases.
So they were talking either around like, well, I’m going to use it to save me time with summarizing notes or editing drafts, or they were talking about lovable.
16:58
So that’s a very specific tool that’s being used at the moment.
I’m hearing a lot about lovable.
I tried to.
Use lovable or lovable.
Is like a product building tool where it kind of like just builds the UI.
It’s so intuitive.
So you can get it to build you an app in a very text friendly way.
17:15
And it’s wow, people are using it like vibe coding, but it builds you an app.
It’s it’s very cool.
I’ve tried to use it.
It’s very interesting.
But then of course, you know, people are are concerned obviously, like what will the implication be around what does it mean for my job?
So we had one person say AI will write curriculum and put me out of a job.
17:32
I’m already looking to switch to another sector.
I think the company will flourish, but I’m not sure about my role.
What will happen to software engineers?
So I think there’s this very real understanding that this can create a positive impact on very specific workflows, but people see themselves during those workflows.
17:52
So what is it going to mean in terms of my job?
But then, you know, some people being more more positive about, yeah, will improve and flourish, but in new ways, which I thought was was interesting, taking up new tools.
The company will flourish.
It’s funny, I think if we did like a mind map, we’d see flourish coming up quite a lot.
18:11
And then again, a lot about, you know, what’s going to happen to my career progression and where’s AI going to help me.
So I think again, what we spoke about these junior roles, what is the impact going to be on the junior roles specifically versus I think people that are in mid Korea where they can transition to different roles.
18:29
You know that the person who’s writing curriculum at the moment and they say that they’re looking for other roles, that’s because they can.
Whereas people that are in the junior roles, I think there’s real concern.
Well, what am I going to do next?
So I saw the same sorts of trends, and then I also saw some interesting trends around this concept of deskilling, where people are encouraging their team to, sure, use AI, but use it in a very intentional way and to not use it for brainstorming and ideation because then it is detracting from their ability to be able to do that and, you know, build on that skill.
19:05
There’s also one person who mentioned that they’ve reached the level of career progression where AI will mostly help them.
And they said, I think for entry level clerk or analyst roles, there will be a lot less options.
I’m not concerned with AI affecting my livelihood, but I’m not sure about the next generation.
And someone also said I think it’s important to still let your brain struggle on its own before using AI.
19:26
And then also someone says they think that their company will be hurt by the way it uses AI because it is taking away from people having their own thoughts and in some cases making sure that the work they sign off on as a human is actually correct.
So there’s some interesting trends in the sense of like maybe don’t use AI for this foundational skill building and use it as kind of a last step in the progression or for other kinds of tasks And you know, keep human in the loop, keep the brain in the loop.
19:55
And maybe for the benefit of the humans as well as like, well, what if the AI is wrong?
Who’s going to check it?
You know, there are some really interesting answers that came up in the next question.
So what do you wish AI could help you with at work?
There’s some funny funny ones.
I thought this was very interesting and started making me worry a lot for when we have humanoid AI more available.
20:18
I don’t know what you thought.
Oh yeah.
So are you thinking about the specific response?
You wish AI could fetch my kids from school so I can work?
That’s the one and the ability to teach complex theory to kids and help them learn new languages.
I’m just like, let’s outsource parenting.
20:35
That’s that’s definitely the one that came up.
So, you know, again, it makes me think if that’s what comes up specifically, it goes back to what we’ve spoken about so many times, which is we want AI to save us time, but you want AI to go and fetch your kids so you can do the more work.
20:53
Right, so time for what?
Yeah.
You had to.
Be a good time.
Yeah.
So it was interesting seeing that kind of response, but then also seeing a lot of responses of people saying basically indicating we want AI to do more complex things, being able to like problem solve, do research, and simply addressing the fact that AI is not at the stage at which it can do that, but they do want it to be applied in that way.
21:18
And then I really liked there is one response, which is what do you wish AI could help you with at work?
Help find emails.
I swear I spend 2 hours a week trying to find old emails.
And if you have Outlook as an e-mail, that is especially difficult in terms of their search feature.
21:35
People all around are always saying how poor Outlook is.
You’re like I type in package and it comes up with 17,000 emails that have nothing to do with package and don’t even have the word package in them.
So I.
Couldn’t relate more, but it is interesting.
21:50
So first of all it is interesting the stuff that they write.
So things like technical problem solving, as you say, finding emails, reliable data crunching, organize my projects and set priorities, productivity, helping manage my schedule.
I guess what’s interesting about all of these things is that this is exactly what AI has been built to do at the moment.
22:13
And some would argue that that is exactly what narrow AI is doing right now and generative AI is doing.
I mean, I think the tools are well suited to this.
So perhaps it’s just that people still need to learn how to use them better.
I mean, other than fetching kids from school.
I can relate that it’s not doing that, but it’s teaching.
22:32
Them very yeah maybe it’s because the companies are essentially saying like here we have the enterprise subscription for ChatGPT or Gemini or Claude have at it versus having the more bespoke you know workflow type organizational applications of AI.
22:49
So maybe it’s just that companies are not at the level of using those, but more so, so instead of like having organizational AI for structure, it’s kind of like individuals are given an LLM to kind of like have at it today.
Maybe that’s kind of just kind of the trend I guess.
23:04
Yeah.
And so I think that’s the big opportunity for companies that want to improve productivity and want to teach people that there is an opportunity there.
Like if if anybody that’s listening here, that’s your game, that’s an opportunity, you know, far be it for us to be encouraging people to make money.
23:22
But like that would be the opportunity for people to kind of improve because these things do do that at the moment.
But to your point, Zach, I’m not sure what they’ll do with their time once they get it back if they are doing this.
But what was interesting again, is when you see like what else do you want to mention?
It really does go back to the lack of regulation, the lack of protection around some of these systems.
23:43
Again, your concern around it, is it going to replace our intellect, concern around our companies pushing this too much and can it disrupt our culture?
Yeah.
And I feel like a good, good quote to end on is the move fast and break things helps no one except perhaps a few people financially at the cost of the 99.99% of the rest of the population.
24:06
So although we have a lot of positive within this survey, we also have a lot of very strong negative and worries about the future and worries about people’s jobs.
And so I mean, we are definitely at this friction point of transition with AI at work.
24:23
And it’s clear that we have a long way to go before it can truly integrate in a way that is successful for a larger group of people, for sure.
Yeah, and it’s not stopping them use it though.
Yeah, there’s that.
I mean, it’s hard when your organization is like, you have to use this, right?
24:40
Yeah.
So I wonder, I wonder if there are any companies that are explicit about not using AI or are putting boundaries between what you can use AI for versus what you can’t use AI for, you know, restricted and bounding.
24:57
No, I imagine there would be depending on client security or system security.
So I think that’s where it would come from.
I don’t think it would come from many other places, but I imagine that it will come from data security places.
Yeah, well, we hope you have enjoyed hearing the results of our survey, and thank you so much for all of you who responded to it.
25:16
And if you have any more anecdotes or things you want to share, please put them in the comments.
We’re always open to continue to hear about your experience with AI at work, especially as it evolves.
So thank you so much for listening to us today for this bonus episode for Series 3.
And this is Bye from Rose.
25:32
And bye from Angie.
See you again.