From GPT to GDP: How Artificial Intelligence is Changing the Workplace
The major sci-fi movies and TV shows of the 1960s helped set the scene: a day would dawn in the 20th century when artificial intelligence would not only exist in the form of robots creating latte art — they would replace everything from taxi drivers to executive assistants.
While some of these visions have yet to become a reality, the workforce has certainly felt the effects of artificial intelligence taking a seat at the table. But not in the way we imagined in sci-fi.
In reality, the onset of AI, and especially the advancement of Large Language Models (LLMs), signals a transformative impact on knowledge workers—the very individuals whose jobs were once deemed irreplaceable by machines. These are the roles that have historically required deep expertise, analytical thinking, and intricate decision-making, from financial analysts and lawyers to researchers and medical professionals.
While manual tasks aren't exempt from automation, it's the roles that demand cognitive skills that are witnessing profound changes. With AI systems capable of digesting vast amounts of data, offering insights, and even drafting coherent texts, the nature of “knowledge work” is undergoing a pivotal shift.
This article will delve into the nuances of this transformation, challenging preconceived notions and exploring the uncharted territories of the AI-driven professional world.
Rethinking AI's Impact on the Workforce
"A lot of stories imagine [automation as] a product like a robot that comes in a box, and you flip it on, and suddenly you have a butler — a perfectly competent and loyal and obedient butler."
According to Seattle science-fiction author Ted Chiang, AI in today's modern world means something much different than what people grew up imagining.
In reality, the onset of AI has fully challenged the preconceived notions employers had related to accelerating and automating work. Instead of following a narrative that suggested the replacement of blue-collar jobs in the service industry, AI has targeted administrative jobs and "knowledge workers." Defined by in-depth cognitive skills, these jobs would have seemed irreplaceable 20 years ago. Now, many argue that more of these jobs are in question.
With the development of large language models (LLMs), AI has advanced further than anyone could anticipate. Tools like ChatGPT have the ability to generate social media strategies, write code, analyze complex data and financial reports, and even write a book. In March of 2023, OpenAI announced that GPT-4 had passed the bar exam with a score of 297, putting it in the 90th percentile of test takers and theoretically on track to practice law in several states.
Many companies are now actively reassessing their workforce in a way most of us would not have predicted a few years ago. Even in 2014, leaders expected a minimal, if not equal replacement of blue-and white-collar jobs. In a Future of the Internet survey, experts like David Clark, a senior research scientist at MIT, hypothesized that automation technology would advance enough to "challenge some of the lower-tier workers," but would still require a similar level of human involvement.
Jerry Michalski, founder of The REXpedition (Relationship Economy eXpedition), a mastermind group founded in 2010, took a different stance on worker displacement:
Automation is Voldemort: the terrifying force nobody is willing to name.
Interestingly, he noted certain "safe zones" (i.e. future career paths that wouldn't be replaced) included local service-based jobs such as gardening, painting, or babysitting, any jobs that required "distant" human effort such as coordination and editing, and positions involving high-level thinking and relationship-building.
While many leading technologists and analysts agreed that artificial intelligence would leave a mark on people's daily lives by 2025, they saw it as unlikely to happen in the short term. Jari Arkko, an Internet Researcher for Ericsson, summed it up best:
There are only 12 years to 2025, some of these technologies will take a long time to deploy in significant scale…
Fast forward a decade, and ChatGPT has over 100 million users and saw 1.6 billion visits to its platform during June 2023; a level of adoption that is largely unprecedented even when compared to other major social media and streaming platforms.
Image Source: Kyle Hailey
With the continued rise of AI adoption, more tasks are on the table to automate. Across the board, employers are at a clear inflection point in redefining what work truly looks like, what jobs appear essential to business function, and which people and processes are on the cutting room floor.
By cutting jobs and replacing tasks with AI, can companies with fewer employees actually produce more?
The Efficiency Quotient: More Output, Fewer Hands
When conducting three case studies across hundreds of similarly skilled business professionals, the Norman Group research firm determined that generative AI (ChatGPT using GPT-3.5) improved employee productivity by 66%, without harming work quality. The study specifically pointed out that the benefit of AI increased for tasks that required more cognitive effort (e.g. customer support vs. coding).
Image Source: Norman Group
While 66% is a significant percentage increase, it doesn't exactly paint a picture of the bigger impact of AI on work. According to the Bureau of Labor Statistics, the average labor productivity growth was 1.4% from 2019 to 2023. This percentage represents the value created by a worker per hour. Over the long term, AI's ability to increase productivity in the double digits holds promise for improving productivity across industries.
Translating percentages to dollars, AI's impact on productivity is estimated to add up to $4.4 trillion annually to the global economy. Nearly 75% of the perceived value would be spread across four main business functions: customer service and support, marketing and sales, software engineering, and research and development.
The online retailer giant, Amazon, is currently implementing AI to identify defective packages at fulfillment centers. Christoph Schwerdtfeger, a software development manager, reports that AI is "three times as effective at identifying damage" compared to a typical warehouse worker. Such findings hint that while AI might enhance productivity and quality, some employers may view AI as a replacement, rather than an assistant.
According to the research outlet Challenger, Gray, and Christmas, there were more than 80,000 layoffs in May 2023, and 5% were attributed to AI. By 2025, AI is expected to replace 85 million jobs. IBM's CEO, Arvind Krishna, shares that the company is making moves to freeze hiring with the expectation that nearly 26,000 non-customer-facing roles would "easily" be replaced by AI in the next five years. BT Group was equally vocal on its plan to cut 55,000 jobs by 2030. Senior leadership shared that their company chatbot, Amy, is already able to deal with most customer inquiries.
At the same time, other perspectives on AI-modified hiring consider the benefits of combining knowledge work with AI, combating the idea that higher-level administrative jobs are suddenly unnecessary. Yann Lecun is one such spokesperson. As the Chief AI Scientist at Meta, Lecun oversees the company's $700 billion in AI projects. He shares the view that AI has the capacity to give employees more time to connect with clients, patients, and customers.
The value of things is going to change, with more value placed on human experience and less to things that are automated.
Another voice arguing that the concerns around knowledge workers are "overblown", is David Luan, an ex-Google and OpenAI employee. He shares his excitement over the "industrialization age" of AI and what that means for workers.
Instead of spending like 30 hours of your week updating Salesforce, you spend 1% of your week asking [software] to just do that for you and you spend 99% of the time talking to customers.
The reality is that AI is already altering the job landscape and future job outlook, beyond a handful of companies, roles, or tasks. However, the consensus is that AI has promise when used correctly, ethically, and in collaboration with existing resources - namely human ingenuity.
Human-Machine Synergy: The Next Evolution of Collaborative Work
ATMs and AI have something in common, points out Jeff Schwartz, Principal at Deloitte Consulting. He notes that since ATMs came to the scene in the 1970s, the fabric of the banking industry hasn't changed all that drastically.
ATMs didn't eliminate the jobs of bank tellers. We've actually seen an increase in the number of bank tellers and in bank branches since ATMs came on the scene.
As ATMs can coexist with bank tellers, AI can coexist with human collaborators. While AI can search for and produce more succinct data and information at a faster pace and greater scale, human decision-making, editing, and critical thinking step in to dictate direction, intention, and implementation. It's a new type of balancing act for teams.
How can businesses create this balance? Understanding what effective human-machine synergy looks like is key, as 19% of U.S.-based companies are categorized as having "barely begun" adopting AI, and some report a lack of trust in it.
From experience, Gretchen Alarcon, Senior Vice President of Employee Workflows Products at ServiceNow, suggests that mitigating overwhelm and mistrust involves companies taking a bottom-up approach and asking employees to engage in the process of deciding exactly how and where AI can streamline existing tasks and processes. She shares that it not only improves AI's efficiency but also incentivizes employees to take the lead in automation changes, rather than feel replaced or "dragged down" by them.
The people who stand to benefit the most from AI are also the ones who will know where AI is needed most.
Another important approach used by the team at NuCompass Mobility is centered around determining whether use cases for automation are actually worth the time to implement them. According to Senior Vice President of Technology and Security, Stephen Chen, it can be easy to say "yes" to every possibility to automate, but it's not always beneficial. Starting small is ideal.
After all, it's the small changes that make the biggest difference in both the day-to-day work and the restructuring of the work environment over time. Small changes add up to big efficiency improvements. And in determining which small changes will make the most impact, AI has another role to play: predictive analysis. Whether it relates to climate control or worker habits, AI is now starting to offer up the potential to optimize workspaces, whether virtual, hybrid, or in-person.
- Automatically scheduling meetings & booking rooms
- Managing equipment and catering
- Optimizing employee workspaces and usage with favorite desk locations, office usage, and commonly used resources
- Managing task assignments based on employee workload
- Improving internal knowledge sharing by connecting employees with resources to upskill
Christian Lehmkuhl, a Design Experience Leader and Design Director at Gensler, shares his vision of an AI-optimized workplace, not only in supporting meeting climate control, making ergonomic adjustments, and scheduling mental wellness breaks, but also in revolutionizing learning and development:
AI-driven personalized learning programs will consider an employee’s current skill level, learning style, and career ambitions to curate a unique learning pathway.
For now, keeping up with the "evolving" AI-assisted job landscape requires preparing current and future employees. The first key is learning to embrace AI, whether in the workforce or education to prepare for a quickly evolving labor market.
Dan Wang, a professor at Columbia Business School, shares that he is actively asking students to use AI tools to be ready for the shifting "battleground of talent" in the hiring process, which he says will require students to consider new and creative ways of harnessing AI for daily tasks. The Columbia University website now includes a resource page detailing "considerations" for how instructors can leverage various AI tools in the classroom, stating that "trying to completely ignore or shut out these tools" does not serve students in the long term for future job prospects.
Redefining Merit in Hiring
While job postings may specify the need for AI familiarity, soft skills remain a staple for job seekers in the age of automation.
Based on an analysis of its Q2 2023 job postings, Freelancer.com suggests that creativity is one trait that will continue to shine for prospective job seekers in the age of AI. Despite generative AI usage tripling, the platform shares that the popularity of data entry and creative writing jobs has actually been on the rise, even though experts point out these are the same jobs that AI is allegedly outmoding. Data processing projects rose by 15%, and creative writing jobs were the fastest-growing job of the quarter, at 58%. Matt Barrie, Chief Executive Officer at Freelancer.com shares his take:
AI can’t replace creativity yet. While workers and businesses are benefiting from productivity gains offered by generative AI, the data suggests that the technology in its current form isn’t able to replace creative work.
While AI can generate creative ideas on a dime, big questions are raised on whether AI can replicate human creativity. One example is the $432,500 sale of The Portrait of Edmond de Belamy. AI is arguably limited based on a lack of personal experience, emotional intelligence, impulsivity, and self-awareness. Even in the cases of creating notable art and work, human ingenuity still acts as the driver.
Another highly-valued skill for employees is problem-solving. Sure, AI can compute on demand. However, algorithms limit the capability to operate in unfamiliar scenarios. Here, AI doesn't have pre-existing data to work from to inform a solution or even to determine which one might be "right" or "best" based on the nuance of a given situation.
And yet another integral human trait needed to balance AI is emotional intelligence. Like empathy, emotional intelligence is uniquely human. The ability to effectively communicate, intuitively read the room, and behave in a way that is culturally sensitive are skills that automated chatbots and machines are currently unable to replicate. Human interaction has yet to find its equal in generative AI and LLMs.
These soft skills aren't just "nice to have" in the hiring process, they are a requirement. Such skills will continue to be valuable as a means to offset the weaknesses of AI that are hugely detrimental in the workplace. Namely, AI bias.
For instance, Amazon called off the use of its AI recruiting tool in 2018 after it discovered it rated women's job applications poorly based on its existing data set, which was found to be predominately male. As a result, AI viewed men as the preferred hires. That's all to say that in the hiring process, the ability to evaluate from a holistic perspective is key, rather than solely relying on AI to evaluate and make final decisions. A human collaborator should still make the final call.
This isn't to say that AI can't be used in the hiring process. When combined with review by a human hiring team, AI-driven assessments can be a powerful way to interview candidates and assess skills and experience in real-time. At Cactus, we’re leveraging AI to simulate sales calls and support requests and provide the hiring team with a concise summary of candidate skills to empower them to make the right hiring decision. Combined with accountability and intentional design, AI holds promise for the future of hiring, while making sure every candidate has the chance to showcase their personality and skills.
The Upskilling Imperative
Despite the previously mentioned 85 million in AI-related job losses, new opportunities are reported to be on the horizon for job seekers. The 2020 Future of Jobs Report estimated that AI will create 97 million jobs by 2025, both in companies looking at the new frontier of AI, and those just starting their journey.
According to a Cognizant survey of 1,000 U.S. senior executives, 20% reported that they were fully utilizing AI, and 61% were just starting to implement it in their processes. Ben Pring, Managing Director at the Center for the Future of Work, says that several executives attempting to use AI are underwhelmed, especially going in with the mindset that AI will "deploy itself." Pring shares that putting in the work is necessary to optimize and see a benefit, and some companies should consider hiring talent that understands how to harness it. In that case, companies could fall further behind, not only in terms of AI use but also in attracting talent.
If you can't get that talent, you can't compete.
In today's global economy, the competition for AI-related jobs and talent is becoming heated. The Adecco Group published its 2020 Global Talent Competitiveness Index. The report not only calls out the fact that AI-based talent is in high demand, but competition presents the opportunity for emerging markets to "leapfrog" into becoming global leaders in both AI and talent. In the report, China, Costa Rica, and Malaysia were identified as potential "talent champions." Other countries like Ghana and India, were called "talent movers" with their ability to grow and retain talent.
The race to find talent is real. Technical students specializing in AI or machine learning at American and Canadian universities currently number in the low thousands, while companies are looking to hire them in the tens of thousands. Universities are quickly adding faculty and programs to accommodate. The University of Montreal reported a maximum enrollment of 200 students for its popular deep-learning course, and Carnegie Mellon is introducing the first undergraduate AI program in the United States. Tom Mitchell, the Dean of the School of Computer Science for Carnegie Mellon, shares that most students will be hired before graduation, where salaries are expected to start in six figures.
Job seekers who are able to upskill and harness AI and machine learning will benefit from the companies that are reportedly willing to shell out big bucks for what Comprehensive.io Founder, Roger Lee, calls "the AI premium."
In a dedicated analysis of tech roles, the salary for a Senior Software Engineer who specializes in AI or machine learning was shown to be 12% higher than a role without specialization. Meanwhile, the average salary for an engineer specializing in AI has already risen by 4% since the beginning of 2023.
The pattern is seen across industries, from entry-level to seasoned management roles. Dropbox shared that it was offering base salaries ranging from $276,300 to $373,800 for AI-heavy roles. Nvidia posted a job for an entry-level research scientist paying in the range of $156,000 to $247,250. The dating app Hinge was hiring a VP of AI with a cool $398,000 annual salary. At one point, Netflix posted an AI Product Manager job with a proposed salary of up to $900,000.
Beyond hiring straight from undergraduate programs, companies can offset the AI knowledge "chasm" by investing in AI training programs that help current employees become more technologically adept. Currently, there are many free options for individuals and teams. Google offers 10 free courses in generative AI, covering everything from large language models to responsible AI and image generation. Microsoft has a nearly three-hour course that walks business professionals through how to adapt and scale processes with AI.
Newer programmers and developers can benefit from courses like Harvard University's free 7-week "Introduction to Programming with Python", which also goes over multiple AI applications. For developers interested in prompt engineering, DeepLearning.AI offers a "free for a limited time" course on how to leverage the OpenAI API and write better prompts, where you can learn more about large language models and have the chance to build your own chatbot. For developers who want to become more efficient in their coding, Udemy offers a course on how to use GitHub Copilot and harness ChatGPT for high-quality pieces of code.
The Disposable Worker
Even with efforts to upskill, there are jobs that won't last the test of time. In the coming decade, the keyword may be "yet". AI can't replace human creativity yet. AI can't replicate empathy yet. Is it more realistic to view all jobs as "temporary" until AI can effectively learn to fulfill them?
People used to say that robots are going to destroy skilled labor. I haven't seen any plumbing robots.
The host of Dirty Jobs and How America Works, Mike Rowe, is adamant that certain skill sets are hard to replicate or fully automate. However, current advancements in AI reflect the ways it has gained a serious foothold in certain industries that many would not have seen coming.
One example of the "temporary job" concept may be seen in Hollywood, as background actors have reported full-length body scans during their time on set. According to various NPR interviews, the scans have become more commonplace as the production teams of major shows and blockbuster movies level up their software skills to generate scenes populated by digitally rendered people.
Image Source: CNET (Lord of the Rings: The Battle of Pelennor Fields)
In the case of production studios, the issue goes beyond simply replacing a worker for one project, as companies are alleged to hold the likeness of individuals to use "for the rest of eternity" without compensation. Time will tell if these “eternal” rights will be upheld.
Similarly, other areas like audiobook narration and dubbing have seen an impact. While those in favor argue that AI helps increase audiobook profitability while reducing barriers to publishing, Emily Lawrence, the cofounder of the Professional Audiobook Narrators Association says the issue goes beyond talking about money; it's about ethics.
If I were to license my voice, and lose all control over how my voice is then used, my voice could potentially be used to voice content that I find morally repulsive.
As is the case with actors in movies, audiobook narrators also face the possibility of entering contracts that don't offer fair compensation when production companies have the freedom to use their voices.
In an adjacent position are voice actors for video games, TV shows, and YouTube projects. In an interview, voice actor Alejandro Graue shared how he had been working on a voiceover project for a self-improvement YouTube channel, which later replaced him with an AI-generated voice. He also shared offers he'd received for gigs related to training AI. By recording 10,000 words, he would receive $52. In some cases, these recording session gigs would be accompanied by contracts that prevented claiming rights to any "voice bank."
From audiobooks and movies to the introduction of self-driving rideshare and TikTok sensation, Waymo, in San Francisco, AI is making its entry into the workforce on multiple fronts, leaving everyone from senior programmers to freelancers and college students to rethink their career trajectory. For people who depend on continued income to survive and support others with each paycheck, the concept of being "disposable" in the new age of AI is not only crushing, it's unethical.
Artificial Intelligence. If you hear those two words again, you are going to throw your f**king computer out the window.
This may be the predominant thought after reading 100+ articles on how AI is replacing every job known to man.
The reality is that the AI conversation is a lot more nuanced than the simple take that AI is here to immediately and permanently replace all jobs. With the introduction of ATMs, electricity, and even the internet, technology has always been a companion to humanity, requiring continuous upkeep, maintenance, and knowledge input.
With respect to AI and its widespread impact, there are still two key issues to consider as we implement and expand upon its capabilities:
- Is it necessary?
- It is ethical?
When business leaders, companies, builders, and creatives come together to answer these questions, a bigger conversation can be had; one that doesn't center around job replacement. Instead, a new vision for the future of work can be created where AI not only creates jobs, it allows displaced workers to be compensated for their work and likeness in the years to come.
If implemented wisely, AI can become a powerful tool in creating better work schedules and processes, leading to a better quality of life. Nearly 100 years ago, the 40-hour workweek was made mainstream. Across the board, studies continue to show that an 8-hour workday limits productivity. Now, AI enters the scene.
The workforce is in a period of reinvention, with the potential to dismantle the idea of a work week. With accountability and intentionality backing AI implementation, it's not a stretch to say that humans and machines can work together to foster better workplaces, better work-life balance, and a refreshed outlook on jobs.
History and sci-fi productions alike hint at the future promise of technology in offering a better quality of life. It's up to us to use AI responsibly and usher in a new era of productivity now that allows humanity to continue to do what we have always done best: create.