Leaders are thinking less about the bean-counting implications of a few cost savings, especially after the pandemic and the Ukraine war have demonstrated that just-in-case may well be more important than just-in-time. Rather, they are focused more on the potential to enhance products and services, as well as individual and organizational performance. There will, of course, be disruption - not that routines have become settled in the past few years - but previous technologies have generally led to more and better jobs for many people and the thinking is that this will do so as well. JL
Rani Molla reports in Re/code:
Half of workers have more than half their tasks exposed to large language models, highest among high-wage jobs: financial analysts, web designers, legal researchers, and journalists. Workers are embracing the technology to do away with drudge work, to be more creative, and up their skills. “AI will not replace you. A person using AI will.” The technology enables them to be better at what they do, likening it to having an incredibly smart assistant at their disposal. The hope is that while the new technology could cause some disruption in what people do, it will ultimately lead to more and better work, like previous technological advancements.A lot of what Conor Grennan does as a dean of students at NYU’s Stern School of Business could be done at least in part by bots. Brainstorming and planning are prime examples of tasks that can be easily handled by generative AI tools like ChatGPT.
But instead of feeling like he could be replaced by AI, Grennan has become an evangelist of this technology and its potential to make work better. He likens the opportunity to work with AI technology right now to finding material wealth.
“It feels like the Gold Rush, like there’s a bunch of people getting to California and seeing little flakes of gold in the river,” he told Vox.
Some of Grennan’s new AI-powered workflow is pretty simple. He drops email chains into ChatGPT or Bing or Bard — he uses them all — and asks it to quickly search for details about a student or deliverables he needs to act on. But he admits that using this technology just for menial tasks would be akin to picking up an iPhone just to use its flashlight. He prefers to use AI to research, brainstorm, and learn.
For example, Grennan might ask an AI to give him five suggestions of a good place to hold an event for 50 students in lower Manhattan on a Tuesday night, or have it reason through how young people in Ohio could help the climate crisis, or have it explain to him what exactly an API is. He keeps asking and refining the questions until he gets good answers, and he’s constantly coming up with new things to ask. Grennan thinks talking with the AI makes him more creative, stokes his sense of wonder, and ultimately makes him better at his job and life.
“Instead of the drudgery of ‘I’ve been given a task, now I’m going to solve that task,’ it’s, ‘I’ve been given a task. What are different ways of looking at it? How can this improve my life? How can I actually get smarter?’” said Grennan. He was recently granted the additional title of head of generative AI at Stern, and is helping develop an AI initiative for Stern’s MBA program so students, faculty, and administrators can become comfortable with AI tools in the workplace.
A recent study by the University of Pennsylvania and OpenAI, the makers of ChatGPT and its more advanced successor GPT-4 — who have a vested interest in hyping their own technology’s capabilities — found that half of workers could have more than half their tasks exposed to large language models, like ChatGPT. Exposure was the highest among high-wage jobs that require degrees and had previously felt relatively safe from the onslaught of technological erasure: financial analysts, web designers, legal researchers, and journalists, among others. While the study said tools like ChatGPT could certainly save those jobs significant time completing tasks, it stopped short of saying those jobs would be fully automated by those technologies. It’s likely, however, that it will change them.
While much has been made about AI’s potential to destroy our jobs, Grennan and other American workers whose tasks overlap with capabilities of software like ChatGPT are embracing the technology to do away with drudge work, to be more creative, and to level up their skills. Marketers are using it to write better copy; programmers are using it to take on projects that were previously out of their league or read code in unfamiliar languages. And it seems like everyone’s now using it to summarize or write emails and boilerplate documents.
Where some see a threat, these workers see possibility. And they hope their mastery of the tools, coupled with their uniquely human skills, will allow them to stay employed even as artificial intelligence gets more and more clever. Back in January, over 40 percent of Americans said they were using generative AI technology at work, and that rate has likely gotten higher. At the same time, about half of Americans think AI could negatively impact the number of jobs in the US.
The truth is we don’t know exactly how artificial intelligence technologies will impact work. The fear is that if tools make existing tasks take hours instead of days, employers might hire fewer employees to get the job done or make their work part-time. The hope is that while the new technology could cause some disruption in what people do, it will ultimately lead to more and better work, much like previous technological advancements, such as the personal computer or even the internet, didn’t spell the end of white-collar jobs.
For now, we know that Americans who’ve chosen to deploy this technology at work seem to like it. They certainly don’t feel like frogs boiling in a pot.
Goodbye to drudgery?
It’s important to remember that even the best jobs have parts of them that suck. And those parts are the first ones on which workers are bringing AI to bear.
Colin McAuliffe, a filmmaker and founder of the production company Zero One Digital Media, has been using the AI software DALL-E to generate images to illustrate business pitch decks. Rather than scroll through pages and pages of stock images, he simply tells the software precisely what he wants — say, a “photograph of a lemon” — and it pops one out.
“It’s something I hated doing or I would make other people do for me,” he said. “And now I do it myself easily, and it’s kind of fun.”
Recently, a client gave him a script that had been written by ChatGPT, and while McAuliffe hasn’t had it write scripts yet, he wants to use AI more often to make shooting schedules and plan trips for his company.
“All that other stuff just takes me away from making videos,” he said.
People say they’re using these tools on tasks they hate, and that allows them to focus on what they love.
Vanessa Bowen, a self-employed product designer, used to dread having to come up with the text that would go inside their app prototypes. If something was off, clients would get hung up on the text rather than critiquing the user interface design of the app, and it could derail the whole interaction. Now, Bowen feeds the AI information about the client and the product, and what type of text box they’re trying to fill; then ChatGPT generates that text.
“It throws out something simple and concise and takes away some of that cognitive load,” they said. That lets Bowen focus on what they really like doing: designing.
“I find that we are stuck in the mundane activities of the day-to-day that could be automated, which then in turn could free up our whole lives,” they added. “It would leave us more time to do other things like be more creative or not work so much.”
It’s a sentiment white-collar workers expressed over and over again, and it’s part of why companies like Microsoft are leaning into AI so heavily in their workplace tools: Not all work is good work. Soon, Microsoft says workers will be able to ask AI-powered tools to make Excel perform complex equations, to have PowerPoint build presentations, and to summarize Outlook emails — all within seconds and by talking to them like you would a person.
Whether that means people will spend that time saved on the parts of their job they really like or whether they’ll simply squander those freed-up hours remains to be seen. It’s also possible that this technology just enables them to make more unnecessary work for everyone else. Take for example, this great cartoon where one person uses AI to make a bullet point into a long email, to which the email’s reader responds by asking the AI to distill the email into a single bullet point.
Leveling up at work
There’s a lot of doom and gloom about what generative AI will mean for computer programmers specifically. Indeed, these tools can often spin up perfectly functional code, using natural language, in an instant, so it’s fair to wonder what that means for the highly paid people who used to do the same thing more slowly.
But the software engineers and developers we spoke to preferred to think of the technology as something that enables them to be better at what they do, likening it to having an incredibly smart assistant or intern at their disposal.
Victor Boutté, a software engineer and tech lead manager at the video hosting company Wistia, says AI tools like GitHub Copilot make him more productive, by suggesting how to complete code he’s started so that he doesn’t have to write the whole thing. Boutté, who works remotely, considers AI tools to be a lot like sitting next to a very smart colleague who also happens to have already read through his code and has limitless time for his questions.
“I’m using it essentially as I would another engineer to bounce ideas off of. It’s helping me flesh through these ideas more deeply, and the feedback is instant,” he said.
AI not only helps Boutté code more quickly, but also elevates what’s possible for him to code in the first place.
“Throughout my career, I’ve never seen a technology as advanced as this. And it gets the creative wheels spinning about what can I use this kind of tech to build,” he said. “It’s inspiring me.”
For independent developer and researcher Simon Willison, AI tools allow him to be more ambitious because he spends less time researching how to figure things out. That means he has more time to try out time-intensive projects he might have previously had to pass on.
Recently, Willison helped his wife with a pottery project. She wanted to see the rate at which a kiln cooled down after being heated in a microwave, in order to estimate its peak temperature while heating. Since she couldn’t put the thermometer in the microwave, after taking the kiln out, she would have to check its temperature over the course of the 90 minutes it took to cool. Instead, Willison asked GPT-4 how to break down a video of the thermometer into 10-second JPEG intervals. He then asked it for commands that would read the temperature from the images and chart it over time. Willison is now thinking about how he could help journalists bring such knowledge to problems like analyzing police body camera footage.
“Normally, when faced with challenges like this, I’d be like, ‘It’s gonna take me an hour to figure this out. Just sit down next to the microwave and write the numbers down,” he said.
Even in less technical and more creative areas, workers are finding that generative AI is able to make them better at what they do. In addition to letting humans spend more time on their creative tasks, generative AI is showing off its own kind of creativity — with the right prompts.
Michael Kaye, director of brand marketing and communications at the dating app OkCupid, has been asking generative AI to come up with in-app matching questions. OkCupid matches people based on how they answer these questions, so coming up with ones that illuminate what’s important to people is incredibly important to how the service functions. At any given time, OKCupid has thousands of these questions available for daters to answer, and Kaye was responsible for creating new ones. For Kaye, this was one of many tasks he does at work, so offloading some of the question creation to AI helped free him up to work on other things. More importantly, he said, the questions that the AI generated — which were based on the simple prompts of “What would you ask on a date?” and “What would you ask on a dating app?” — were actually very good.
The first 10 AI-generated questions Kaye ended up adding to the dating service included “How do you balance your own needs with the needs of your partner in a relationship?” “What do you value most in a partner?” and “Are you a morning or night person?” So far, they’ve been popular, with users responding to them more than 675,000 times since the end of January.
“They’re high-quality, especially given how generic the prompt was,” Kaye said. “They might sound surface-level, but I think those are things that really help connect people.”
He plans to add more ChatGPT questions using more specific prompts every month this year.
Why AI probably won’t take our jobs
While everyone we spoke to understood that generative AI might be disruptive to some jobs, no one felt it was a real danger to theirs.
A common refrain was a version of a tweet from machine learning engineer Santiago Valdarrama that said, “AI will not replace you. A person using AI will.” In other words, they felt that their mastery over generative AI tools would give them a leg up, even if that generative AI made some of what they’re paid to do obsolete.
“When I first started using GPT-4, [losing my job] was my first concern. It’s very natural to feel threatened by new technology, especially technology that’s really good at what you do,” said Stephanie Yamkovenko, a group manager for the digital marketing team at Khan Academy, an education nonprofit that has partnered with GPT-4 maker OpenAI since September.
“But as I’ve used it more, I’ve realized that it’s going to be a skill that’s going to be in high demand for writing and editing in the future,” she added.
Yamkovenko recently was able to work on a much more robust product launch than what her small team normally would have been able to do. She used ChatGPT to write a greater volume of social media copy, which in turn snagged the company 10 times the traffic it would normally get.
Others were sanguine that their skills, now boosted by their agility with generative AI, are eternally in demand.
Somnath Banerjee, VP of data science at the early-stage investment firm Clear Ventures, said there will always be an abundance of work for engineers like himself. He’s been using AI to code projects more quickly and to feel more confident in his email writing as a non-native speaker.
“They will not say, ‘I paid him for two weeks and you did it in two days,’ because there’s always two years of work waiting for you,” he said.
None of this, of course, is to say that work will be the same. The acceptance of using tools like ChatGPT will be the first of many changes. Even if they do keep their jobs, what these white-collar workers do and how they do it will likely be different if generative AI technology becomes widely used. That will be a loss to many who like their craft as is.
Take, for example, this 3D mobile games artist who lamented on Reddit recently: “My job is different now since Midjourney v5 came out last week. I am not an artist anymore.”
The poster added, “All I do is prompting, photoshopping and implementing good looking pictures. The reason I [wanted] to be a 3D artist in the first place is gone. I wanted to create form In 3D space, sculpt, create. With my own creativity. With my own hands.”
Those who’ve embraced generative AI and who are less tied to the specific tasks of their work will likely have an easier time adjusting to a world of work that’s changing in front of their fingertips.
0 comments:
Post a Comment