AI's Quiet Tsunami, Part 3: Next, AI & (Your) Work

I hope you’ve read part 1, AI’s Quiet Tsunami, and part 2, Challenges in a World Fueled By AI, as well as My Conversation With AI. (You may also want to scan “Next, Cue the Robots.”) Here I’m going to focus specifically on AI and (most) work. Since I’ve been getting so many questions about education, that’s where the fourth in the series will focus. And the fifth will be on what we can — and must — do going forward.

Net Next: Both of these things are true… AI tools are unleashing a pre-Cambrian era of innovation and creativity, with seemingly unlimited new opportunities. And these tools will have a tremendously disruptive impact on a lot of existing work.

Key Points About AI+Work

At a talk in Budapest a few weeks ago to a group of brilliant students at the University of Technology and Science, during the launch of The Next Rules of Work translated into Hungarian, a student asked me, “What will happen to people and work in the future?” — a common question in late 2019, until a virus came along.

In response, I want to offer four key thoughts related to AI and work.

• These tools are empowering many people to do more created work. Human creativity just got an electric jolt. If you use text or images or video in your work, activities ranging from essay writing to presentation design are being transformed. But it’s not creative work, performed by humans: It’s created work, which you can build from. Still, if we all use them, then…
• A significant amount of work is likely to be disrupted by these tools. The software will automate a lot of tasks that people used to perform, many of them activities that seemed to be uniquely human work. As always, that’s a threat to some, but an opportunity for others. Yet we can’t predict the specific impacts, because…
• Robots and software don’t (have to) take human jobs. Technology simply automates tasks and processes. It’s a human’s decision if those automated tasks add up to a job that goes away. And we can make different decisions. Including…
• We can — we must — help everyone affected to adapt. We need to empower everyone affected to learn how to use these tools to their advantage. We can do that in part by making the tools far more human-centric, including making them easier to access, learn, use, and control. And we can each make a commitment to keep providing meaningful, well-paid work for humans.

Work That’s Empowered: You Become an AI Whisperer

It’s been tremendously inspiring to watch creative people around the world hacking the limits of the new crop of tools. (As my friend and famed hacker Pablos Holman says, “A user asks, what does it do? A hacker asks, what can I make it do?”)

A legion of experimenters has made the new generation of tools write email messages, generate Wikipedia entries, check medical symptoms and receive patient care recommendations, write code and fix bugs (even at Amazon), get suggestions for product management, get ideas for art, decoration, and party themes, find out what’s missing from an employee handbook, write real estate listings (and other “proptech” hacks), write cover letters and resumes and (slightly) improve chances for getting in the door for a job, prepare sales collateral and generate copy for paid search ads, build websites in 30 seconds, boost productivity, create a text adventure game, sift through reading material and generate notes, build workout plans, write journal entries and get feedback, get life advice, get ideas for photography, and win a photography contest with a generated image. (Just be sure to give back the prize money).?

You can even auto-generate your own Eminem rap on AI, using ChatGPT to write it and uberduck to deepfake the voice.?

Just don’t expect the tools to create “perfect” text or code or images or songs (yet). The Atlantic’s Derek Thompson accurately says these tools today “mimic creativity.” Put another way, these tools are doing more created work, so you can do more creative work. A lot of AI-generated content still requires a professional to review and edit the output. So to get the best output, you’ll need to be “an AI Whisperer,” such as a Prompt Engineer who can tweak instructions to AI programs to get the best output.?

These tools are doing more created work, so you —

as an AI Whisperer — can do more creative work.

Similar job titles of the future may include Chatbot Developer, and Language Model Trainer, Scripter, Translator, Researcher, Journalist, and Educator — wrangling the tools so they create the results that humans want.

As Marco Andre at Novartis said, “Coders will become prompters. Customers will become creators. Marketers will become orchestrators.”

Work That’s Impacted: If You Don’t Whisper to AI

Of course, if each of us can do these AI-fueled tasks more easily, what becomes of those we used to pay to do these tasks? It’s a little bit of hyperbole to ask, Is no career safe? If you deliver food to a restaurant table, or scan and bag groceries in a store, your work probably won’t be affected very much by the new AI tools. (Robots, yes. AI software, not yet.) In How ChatGPT Will Destabilize Work, The Atlantic’s Annie Lowery wrote, “No single technology in modern memory has caused mass job loss among highly educated workers.” (Yet.) Longtime tech journalist Rafe Needleman suggests that GenAI “...will impact your job, but it won’t kill it.” That, of course, depends on the decisions of those who pay you.

The basics of impacted work are:

• Work is human skills applied to tasks to solve problems.
• AI tools perform tasks.
• Tasks add up to a job.
• A human decides if a job goes away.

Automation has always first focused on the “lower left” of the task landscape — repetitive work that doesn’t require much creative thought. Humans remain great at the “upper right” — unique tasks requiring creative problem-solving. Think of the new AI tools as “the created middle” — not truly creative, but good enough if you don’t need high standards. That doesn’t automatically mean that jobs have to go away. But if you’re a freelancer creator, and your clients stop calling you because they’re using AI tools, you’re the canary in the coal mine for AI’s quiet tsunami, just as journalists like me were impacted in the Internet tsunami.

The tech analyst Benedict Evans called these tools “infinite interns.” (Often silent interns: though research groups are banning the practice of crediting AI as a co-author — sigh of relief — some researchers are likely to use AI tools without mentioning it.) And like 10-year old interns, you’ll need to ride herd on them — or, even better, pay creators to ride herd on them.?

There is a lot of speculation about the impact on jobs. Here are some of the work roles that are likely to be affected by AI tools in the “created middle.” (Don't forget that some of these were already heavily impacted by automation.)

• Data entry and data processing. As AI gets better at recognizing source material like the spoken word, more of these “lower left” tasks get automated.
• Customer service and customer support. Repetitive requests get automated first, and humans answer more sophisticated questions. (Then, humans train the AI software to do that.)?
• “First draft” Report Writing. Is this the end of writing? You may not use these tools for an entire essay, but a first draft or two that you can personalize.
• “First draft” Text Translation & Transcription. There already are quite a few translation engines out there, like Google Translate. But a simplified interface like ChatGPT makes it translate to do pages of text, then a practiced translator or writer does the final edit.
• “First draft” Storyboard Art will still often require a competent artist, but the AI tools should be good enough to get initial ideas across.
• Programming. Though many agree no tool can (yet) generate sophisticated code, Google tested ChatGPT — and if “hired” could make over $180,000US a year. And OpenAI is already working on a next-generation code generator.
• Voiceover Acting. With the stunning effectiveness of deepfake tools, voiceover actors are being asked to sign away rights to their own voices to be used for auto-generated content.
• Music. AI music generators today can help musicians generate additional tracks and alternatives. For the tone-deaf like me, I'd add a basic soundtrack to simple content, but hire a musician for anything of quality.
• Journalism. A field already ravaged by the Internet and online ad markets is about to get another reset, as formulaic “lower left” articles like sports reporting are increasingly auto-generated. Despite fallout about CNET’s lack of transparency for its AI-generated articles, Buzzfeed’s stock doubled on its announcement to do the same.
• Art and photography. It’s possible to spot AI-generated art, but probably not for long. Shutterstock announced a feature that autogenerates images using OpenAI’s DALL-E2, and immediately licenses the image, no humans required. Does that mean buyers will pay less for human-created art and photos? Will only the most-talented (or most marketable) artists and photographers continue to have a livelihood??

Just remember that rumors of the death of jobs have included bank tellers, grocery-store checkout clerks, truck drivers, and cab drivers, each of whom has been predicted to face a jobpocalypse. Financial advisors continue to work for high-touch clients. And while there are many coaching and therapy bots, so far they are only useful for simple issues.

So far, the main result of automation software is that people often get paid less, and have worse jobs. According to a recent MIT study, the kind of automation that duplicates human work inevitably results in lower wages for workers, and the robotification of existing work.?(Though in factories, robots are clearly used to replace human jobs.) You can look at my 2014 piece on Unbundling Media to see how job disruption happened to journalism, and at my 2017 piece on Unbundling Work to see the dynamics for other work roles.?

AI Tools Do Not Need to Take Human Work — If We Help Everyone Adapt

Will the new AI tools inevitably result in a jobpocalypse future? Or will people be able to continually adapt? In The Next Rules of Work, I suggest three possible futures: 1) Lots of robots but few jobs. 2) Lots of jobs even with lots of robots. 3) Lots of available jobs, lots of robots — and lots of unemployed, untrained, and unprepared workers. Many people worry about #1, but right now we have #2, and these new tools could accelerate the timeline for #3.

Of course, new technologies have always transformed work, from the seed drill to the knitting frame. But new technologies alone do not take away work: It is the business models of the owners of technology that disrupt workers. (I’ll cover more about this in #5, Preparing for Our AI Future.)

As has happened throughout history, one of the common drivers of automation is when employers can’t find the skills they need. In December ‘22, when tech companies were laying off tens of thousands of workers, there were 11 million open jobs. That’s a lot of demand, and if employers can’t find the skilled workers they need (such as in the hospitality industry), many as usual will attempt to leverage technology to fill the gap.

If we do this wrong, a small number of software engineers will generate huge markets for their products, with zero concern about the impact on human work. But if we do this right, we will actually co-create the “next” rules, together, so that the vast majority of people will benefit.?

There is no law of nature that requires technology to destroy jobs. Job loss comes through a series of decisions by humans, especially when organizations refuse to train workers, and simply discard those who are no longer needed. The organizations that make the commitment to leverage the new AI tools, and to continually train workers in their use, are the ones who will have the workers they need to solve the problems of today and tomorrow.

If employers embrace these new tools, they will provide workers with a baseline ability to create basic content, allowing them to concentrate their brainpower to perform higher-level critical thinking and analysis, just as Plato would have hoped. Of course, that will require AI Literacy — which some call AIQ — giving everyone a basic understanding of these tools. And it will require the developers of the tools to make them free or inexpensive, accessible, transparent, and accurate.

What could you do Next?

• If you are an employer, start internal task forces (we call these NEXTlabs), now, to make sure you deeply understand the ways these tools can benefit your workers, and train workers in the use of the tools. (SAG-AFTRA, of all groups, hosted a good series of sessions on Labor and AI.)
• If you lead a team, and have the potential to have these tools used in your team’s work, commit to ensuring that they will indeed be freed up to do other work, and not simply become relegated to doing boring work, or lose their jobs
• Assemble cohorts of people (again, NEXTlabs) to help each other learn and adapt. If you’re a creator, join a creator network, like this one.
• Get infected by possibilities. Watch this video on ChatGPT and your career.

Thanks to Herman Gyr, Ph.D. , Shelly Palmer , and MELANIE CHARTOFF for pointers to news items.

-gB Gary A. Bolles

I’m the author of?The Next Rules of Work: The mindset, skillset, and toolset to lead your organization through uncertainty; the adjunct Chair for the Future of Work for?Singularity Group; have over 1.3 million learners for my courses on?LinkedIn Learning; a partner in the consulting firm?Charrette LLC; a co-founder of?eParachute.com; an original founder of?SoCap Global, and former editorial director of 6 tech magazines. Learn more at?gbolles.com