How I Learned to Stop Worrying and Love AI
Artificial Intelligence will change the world and your work. But how it changes is up to you. Here are a few things I’ve learned working with AI and ML that may help you calm your nerves and step bravely into a rapidly evolving future.
The first thing I did when given access to unparalleled machine intelligence? I used GPT-3 to create clues in rhymed couplet format for a scavenger hunt in my Brooklyn neighborhood for my son’s 13th birthday party. And then asked for cocktail suggestions based on the meager contents of my liquor cabinet. Neither changed the course of human history. Both were a raging success.
Why was my liquor cabinet so depleted, you ask?
Because 2022 and 2023 have been the best of times and the worst of times for the tech industry. A dramatic loss in jobs as companies adjust to the inflationary macroeconomic environment and cut costs has come at the same time as a Cambrian explosion in technology innovation.
A ton of powerful breakthroughs have landed at once, powered by machine learning and artificial intelligence. Speech recognition, natural language processing, computer vision, Large Language Models, oh my! These are not brand new capabilities but they have been unleashed by an unprecedented capability for processing, coupled with vast troves of data and the ability of machines to make sense of a larger number of inputs in the form of human expressions such as doodles, scribbles, and images.
And perhaps one of the biggest unlocks when it comes to driving innovation: access to these AI capabilities has been democratized via an intuitive conversational interface; no math is required! When OpenAI’s ChatGPT exploded onto the scene in November 2022, it reached 100M active users in just two months, setting the record for the fastest-growing application ever. Regular folks can now “speak machine” and conjure all manner of ‘original’ works.
OpenAI’s ChatGPT reached 100M active users in just two months, setting the record for the fastest-growing application ever
As a result of this confluence of technical advancements, we are both at a genuine inflection point and the AI hype cycle is at its frothy peak. Projections of its massive economic potential and societal impact have become daily fodder for mainstream media, global think tanks, management consultancies, and LinkedIn AI influencers carpet-bombing your feed with content offering you the formula for how to 10x your productivity/revenue/followers/etc.
Inside organizations, people are feeling new pressure to harness the power of AI. While a large number of the >400K professionals impacted by tech layoffs over the past two years are wondering if AI was the job killer; especially those in ‘creative’ disciplines such as design, marketing, writing, and software engineering who see the marvels of generative AI being touted as a disruptor and threat to their craft.
Cue existential dread and a collective crisis of confidence. Where do I fit in? What is my role? What will happen to this craft I’ve spent so much time and energy learning? Am I going to be replaced? When, specifically, will some kind of AI supplant humankind as the dominant intelligent species on the planet making “the very existence of the species of man a condition precarious and full of terror”? (shout out to Mary Shelley, the O.G. ‘The AI Gal’ who warned of the risks when imbuing human vitality into inanimate objects)
I don’t have the answers to these questions. And I don’t have a magic formula for success.
However, I do have a few decades of experience honing several crafts centered on applying emerging technologies. As well as a few years of experience leading a portfolio of work at McKinsey & Company focused on evolving that firm’s operations with innovative applications of AI and on growing its capability to scale the impact of these technologies to its knowledge-intensive work.
From this, I can draw a few observations about how to navigate the current gestalt and prepare yourself for what comes next…
Accept that these recent advances in Artificial intelligence represent—at an absolute minimum—a technology-driven disruption equivalent to the advent of the web and of mobile computing.
There will be exponential growth and transformation, with companies disrupting existing business models and ushering in an era of hyper-innovation. There will be lots of fools gold and lots of snake-oil salesmen. It is too big to ignore, yet its future is far from certain. As Bill Gates wrote, “we always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten.”
It seems clear already that the way we work, learn, and entertain ourselves will change. After all, AI has already been doing that behind the scenes for the past decade, with ML models powering large swaths of modern software and orchestrating the experiences we have mediated by digital technologies. Now, the great leap forward triggered by advances in Large Language Models has unlocked AI’s ability to make sense of a wider range of inputs and generate a wider range of human-like content, unlocking new opportunities to augment human creativity and improve productivity.
Understand that—at least in the near term—AI will take over tasks, not jobs. But the nature of our work will be reshaped significantly, for some more than others.
If we look at the whole arc of human history, the one thing that’s been consistent is that more and more work gets automated via technology. LLMs are just the latest instance. History also tells us that worker displacement from automation has been offset by the creation of new jobs. The telegraph put a lot of carrier pigeons out of work but introduced telegraph operator as a whole new occupation.
Portrait of a victim of automation (image created using Stable Diffusion)
But for the majority of the workforce, technological advancements tend to automate tasks, not entire occupations. And AI is no different.
According to research from McKinsey, by 2030 only 5% of occupations can be completely replaced by AI, whereas fully half of today’s work tasks could be automated during that same period of time. The exact shape of the workforce and accounting of how many jobs will be affected by AI is different depending on who you ask. But the historical precedent is directionally clear.
Regardless of your industry and discipline, the nature of your work will change. For some more than others. For most, AI will be a tool that helps them get more done. In other scenarios, the technology will replace labor rather than complement it. Or turn a job from one that requires special skills to one that doesn’t.
Rather than thinking of AI as automating jobs, it’s more useful to think about jobs as collections of tasks and to analyze AI’s ability to augment or automate individual tasks.
Accept that how you understand and apply your craft must evolve to remain relevant in the workforce.
Whether you’re a financial analyst, medical researcher, screenwriter, or software engineer, this should be a time for reflection and experimentation. Where can I delegate rote, repetitive tasks to AI? How can AI be a tool to augment my ability to do parts of my job? And what new skills do I need to develop that complement AI systems?
Answering these questions should not be a paper exercise. And those who wait passively for the answer to emerge are at a significant disadvantage. As Scott Galloway, entrepreneur and professor of Marketing at NYU’s Stern School of Business, puts it:
“You shouldn’t be scared of losing your job to AI, you should be scared of losing your job to someone who understands AI”
How will you integrate AI to enhance your productivity, creativity, and communication? (image created using DALL-E)
To become someone who not only understands AI but develops a distinctive point of view on how to actively apply these technologies in ethical and valuable ways, I would suggest a few starting points:
1) Build your AI literacy. In order to take advantage of AI’s evolving capabilities, you need to understand how it works, what it can be used for, and how it can be used best to enhance performance and achieve desirable outcomes. You don’t need to become a data scientist or ML engineer, but you do need to invest in growing your technical acumen; learning foundational concepts and technologies so that you have an understanding of how AI systems work. Applied AI also comes with a host of practical, legal, security, and ethical concerns that need to be considered so AI literacy is an important facet of digital citizenship, our ability to engage competently and positively with digital technologies and participate actively in shaping how it changes our work and communities. And if you are a knowledge worker of any stripe, it is a non-negotiable job requirement.
2) Make time to tinker with AI tools and technologies. The only way to figure out how useful AI might be is to use it. To make effective use of this technology, you must learn how to play to its strengths and avoid its weaknesses. AI has a remarkable ability to do certain things better and faster than humans (e.g. identify patterns and illuminate opportunities), and it has real limitations and risks (e.g. hallucinations, bias). But until you put them to work, you won’t know the best ways to use generative AI tools like ChatGPT or Midjourney, or the conditions under which they fail to produce helpful output.
For example, my nascent skill as a prompt engineer and naive delegation to an ‘overconfident bullshit machine’ almost spelled disaster for that scavenger hunt I co-created with ChatGPT. Only a trial run the night before my son’s birthday party revealed that one quest in the hunt was a dead end, requiring a visit to a local business that had permanently closed after September of 2021, the cut-off for training data used to inform GPT-3, the underlying foundation model.
Christoph Niemann’s cover for The New Yorker accurately captures the unpredictable outcomes of applying the current generation of generative AI tools; you must learn to play to their strengths and weaknesses, which vary depending on the underlying foundation model
In the field of industrial design, there is a concept called ‘material exploration’ which prompts designers to consciously explore materials and to draw on that exploration for ways the material could be used in a solution. Through this process of prototyping, testing, and evaluating the results you gain an understanding of the affordances, possibilities, and constraints inherent in working with a particular medium. This type of experiential learning deepens your grasp of concepts, engages your reflective observation skills, and stimulates creative problem-solving. All of which accelerates progress towards productive application.
With emerging technologies, novelty often precedes utility and playful tinkering and exploration are the best ways to begin to generate insight about the potential for more purposeful and productive uses. This is true at an individual level. But also as an organization. If you’re a business leader, now is not too late to help your workforce start to explore the potential of AI.
3. Actively reflect on your craft and where AI might fit. Now is a time to examine What do I do and how do I do it? What do I do that creates the most value? What (tasks) should be automated or augmented with AI? (note: should, not can)
As someone on the frontlines of the ascendancy of artificial intelligence, Chess Grandmaster Garry Kasparov offers some unique insight into how to successfully integrate these technologies in order to enhance your performance.
After losing to IBM’s Deep Blue in 1997, Kasparov began to experiment with how the introduction of AI co-pilots changed chess players’ competitive advantage. What he discovered was that having the best players and the best program was less a predictor of success than having a really good process. As he expressed it: “Weak human + machine + better process is superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process”.
More recently, a 2023 study of AI-augmented knowledge work, has reached similar conclusions. 758 consultants at BCG, a global management consultancy, were randomly assigned to one of three groups: no AI access, GPT-4 access, or GPT-4 access + an overview of prompt engineering.
For each one of a set of 18 realistic consulting tasks easily performed by AI, the consultants with access to AI were significantly more productive and produced significantly higher-quality results. Consultants using AI finished 12.2% more tasks on average, completed tasks 25.1% more quickly, and produced 40% higher quality results than those without.
The less-skilled consultants (based on performance benchmarking at the beginning of the study) experienced a more significant lift from AI augmentation (+43%), than those who were more skilled (+17%)
Interestingly, this study also explored what happens when AI is applied to a task for which it is not suited. For those tasks deemed beyond “the technological frontier” of current LLM capabilities, the consultants using AI were 19% less likely to produce correct solutions compared to those without AI
In other words, and as Kasparov presaged, knowing when and where to apply AI can significantly enhance your performance. And using it inappropriately can actually put you at a disadvantage, leading you to worse outcomes than you would have otherwise achieved.
4. Commit to continuous learning as a mindset and habit. Kasparov is an inspiration. Publicly humbled by an earlier generation of this technology, he set out to understand what happened. Not viewing AI as an existential threat, but instead with curiosity. How can human intelligence be augmented by this tool? How does my unparalleled and widely recognized expertise in this domain evolve as a result of this new development?
“In the beginner’s mind there are many possibilities, but in the expert’s mind there are few. ” - Shunryu Suzuki (image created using DALL-E)
Kasparov’s response highlights something that will be important going forward in the fast-changing and unpredictable frontier of AI-driven change: the need to adopt the ‘beginner’s mind”, letting go of your preconceptions and having an attitude of openness about how your craft, occupation, and professional identity should evolve.
“Be aware of the assumption that the way you work is the best way simply because its the way you’ve done it before”
This has always been true, but more so in the future as lifespans extend and societal change accelerates. There is no done. Mastery is a moving target. And you’ll need to make time to become your future AI-augmented self. In part, because AI is not a set-it-and-forget-it technology. These tools are evolving rapidly so you can’t learn new skills once and be done. You will need to continually revisit your workflow and continually refresh your learning about AI’s capabilities, emerging best practices, and the latest regulations.
But don’t be daunted! You are not too late. There are no experts. Sam Altman, CEO of OpenAI, does not know the implications of AI for your job or your company. While we generally understand what AI can do quite well and there’s lots being published about the potential impact of use cases in various industries, there is much to be figured out in practice, especially when it comes to applying generative AI.
The inner-workings of LLMs are an enigma, even to those creating them:
Interaction paradigms are still nascent ($5 says a text prompt in a chat dialogue is not the end state)
Outcomes are quite unpredictable
AI has not so much come of age as reached an awkward adolescence. And just like my 13-year-old, now is the time to spend an inordinate amount of time in front of the mirror and try on a few different looks. Your future self will thank you for it.