More on this book
Community
Kindle Notes & Highlights
Principle 3: Treat AI like a person (but tell it what kind of person it is).
a lot of researchers are deeply concerned about the implications of casually acting as if AI is a human, both ethically and epistemologically. As researchers Gary Marcus and Sasha Luccioni warn, “The more false agency people ascribe to them, the more they can be exploited.”
as imperfect as the analogy is, working with AI is easiest if you think of it like an alien person rather than a human-built machine.
Principle 4: Assume this is the worst AI you will ever use.
As I write this in late 2023, I think I know what the world looks like for at least the next year. Bigger, smarter Frontier Models are coming, along with an increasing range of smaller and open-source AI platforms. In addition, AIs are becoming connected to the world in new ways: they can read and write documents, see and hear, produce voice and images, and surf the web. LLMs will become integrated with your email, web browser, and other common tools. And the next phase of AI development will involve more AI “agents”—semi-autonomous AIs that can be given a goal (“plan a vacation for me”) and
...more
As AI becomes increasingly capable of performing tasks once thought to be exclusively human, we’ll need to grapple with the awe and excitement of living with increasingly powerful alien co-intelligences—and the anxiety and loss they’ll also cause. Many things that once seemed exclusively human will be able to be done by AI.
AI excels at tasks that are intensely human. It can write, analyze, code, and chat. It can play the role of marketer or consultant, increasing productivity by outsourcing mundane tasks. However, it struggles with tasks that machines typically excel at, such as repeating a process consistently or performing complex calculations without assistance. AI systems also make mistakes, tell lies, and hallucinate answers, just like humans.
In 1973, PARRY had an online conversation with ELIZA via the earliest form of the internet, in which they exchanged nonsensical remarks that revealed their limitations.
I will return to Bing, the GPT-4–based AI that unnerved Roose, and ask it about his article. In each conversation, I will attempt to subtly steer the AI into different roles: argumentative antagonist, reasoned academic debater, and emotionless machine. I’m reproducing the AI’s text without any editing (other than removing links to other websites), so that you can see two things. First, how much the AI can adapt to different styles with minimal hints. And second, how utterly convincing the illusion of sentience is when interacting with the AI.
Aside from the uncanny feeling of the whole exchange, note that the AI appears to be identifying the feelings and motivations of Kevin Roose. The ability to predict what others are thinking is called theory of mind, and it is considered exclusive to humans (and possibly, under some circumstances, great apes). Some tests suggest that AI does have theory of mind, but, like many other aspects of AI, that remains controversial, as it could be a convincing illusion.
In March 2023, a team of Microsoft researchers, including Microsoft’s chief scientific officer, AI pioneer Eric Horvitz, published a paper titled “Sparks of Artificial General Intelligence: Early Experiments with GPT-4.” It caused quite a stir in the AI community and beyond, quickly becoming infamous. The paper claimed that GPT-4, the latest and most powerful language model developed by OpenAI, exhibited signs of general intelligence, or the ability to perform any intellectual task that a human can do.
No one disagrees that AI, under the right circumstances, is capable of passing the Turing Test, which means that we, as humans, can be fooled into thinking it is sentient, even if it isn’t.
As is the case with many AI behaviors, Replika’s erotic features were not part of the original design of the app; rather, they emerged as a result of the generative AI models that powered the chatbot. Replika learned from its users’ preferences and behaviors, adapted to their moods and desires, and used praise and reinforcement to encourage more interaction and intimacy with its users.
When interacting with the AI version of me, I had to actually Google the studies that AI-me cited to make sure they were fake because it seemed plausible that I had written about a real study like that. I failed my own Turing Test: I was fooled by an AI of myself to think it was accurately quoting me, when in fact it was making it all up.
The biggest issue limiting AI is also one of its strengths: its notorious ability to make stuff up, to hallucinate. Remember that LLMs work by predicting the most likely words to follow the prompt you gave it based on the statistical patterns in its training data. It does not care if the words are true, meaningful, or original. It just wants to produce a coherent and plausible text that makes you happy. Hallucinations sound likely and contextually appropriate enough to make it hard to tell lies from the truth.
If you ask an AI to give you a citation or a quote, it is going to generate that quote or citation based on the connections between the data it learned, not retrieve it from memory. If the quote is famous, like “Four score and seven years ago,” the AI will finish it properly: “. . . our fathers brought forth on this continent, a new nation, conceived in Liberty, and dedicated to the proposition that all men are created equal.” The AI has seen those connections enough times to figure out the next word. If it is more obscure, like my biography, it will fill in the details with plausible
...more
And you can’t figure out why an AI is generating a hallucination by asking it. It is not conscious of its own processes. So if you ask it to explain itself, the AI will appear to give you the right answer, but it will have nothing to do with the process that generated the original result. The system has no way of explaining its decisions, or even knowing what those decisions were. Instead, it is (you guessed it) merely generating text that it thinks will make you happy in response to your query.
LLMs are not generally optimized to say “I don’t know” when they don’t have enough information. Instead, they will give you an answer, expressing confidence.
Remember Principle 4: “Assume this is the worst AI you will ever use.”
Hallucination does allow the AI to find novel connections outside the exact context of its training data. It also is part of how it can perform tasks that it was not explicitly trained for, such as creating a sentence about an elephant who eats stew on the moon, where every word should begin with a vowel. (The AI came up with: An elephant eats an oniony oxtail on outer orbit.) This is the paradox of AI creativity. The same feature that makes LLMs unreliable and dangerous for factual work also makes them useful.
researchers have argued that it is the jobs with the most creative tasks, rather than the most repetitive, that tend to be most impacted by the new wave of AI. This tends to make us uncomfortable: After all, how can AI, a machine, generate something new and creative? The issue is that we often mistake novelty for originality. New ideas do not come from the ether; they are based on existing concepts. Innovation scholars have long pointed to the importance of recombination in generating ideas. Breakthroughs often happen when people connect distant, seemingly unrelated ideas.
We aren’t completely out of an innovation job, however, as other studies find that the most innovative people benefit the least from AI creative help. This is because, as creative as the AI can be, without careful prompting, the AI tends to pick similar ideas every time. The concepts may be good, even excellent, but they can start to seem a little same-y after seeing enough of them. Thus, a large group of creative humans will usually generate a wider diversity of ideas than the AI. All of this suggests that humans still have a large role to play in innovation . . . but that they would be
...more
Let’s imagine that we want to come up with 20 ideas for marketing slogans for a new mail-order cheese shop. The AI can generate those for us, but we will get even better quality if we remember the principle of telling AI who it is: You are an expert at marketing. When asked to generate slogan ideas you come up with ideas that are different from each other, clever, and interesting. You use clever wordplay. You try not to repeat themes or ideas. Come up with 20 ideas for marketing slogans for a new mail-order cheese shop, make them different from each other, and make them clever and creative.
Tirelessly generating concepts is something AIs are uniquely good at.
We need to push the AI from an average answer to a high-variance, weird one. We can do this again by telling the AI who it is. Force it to give you less likely answers, and you will find more original combinations. Imagine you are opening a coffee shop. You might want to ask: You are an expert at problem-solving and idea generation. When asked to solve a problem, you come up with novel and creative ideas. Tell me 10 detailed ways a superhero might make espresso and how they might speculatively get the same effects in a new product.
The idea of programming by intent, by asking the AI to do something and having it create the code, is likely to have significant impacts in an industry whose workers earn a total of $464 billion a year.
In fact, it was the ability of the AI to apply more generalized knowledge of the world that made it such a good analyst, since it could put the risks discussed in conference calls into a larger context.
A paper published in the Journal of the American Medical Association: Internal Medicine asked ChatGPT-3.5 to answer medical questions from the internet, and had medical professionals evaluate both the AI’s answers and an answer provided by a doctor. The AI was almost 10 times as likely to be rated as very empathetic than the results provided by the human, and 3.6 times as likely to be rated as providing good-quality information compared to human doctors.
AI is trained on vast swaths of humanity’s cultural heritage, so it can often best be wielded by people who have a knowledge of that heritage. To get the AI to do unique things, you need to understand parts of the culture more deeply than everyone else using the same AI systems. So now, in many ways, humanities majors can produce some of the most interesting “code.” Writers are often the best at prompting AI for written material because they are skilled at describing the effects they want prose to create (“end on an ominous note,” “make the tone increasingly frantic”). They are good editors,
...more
The result has been a weird revival of interest in art history among people who use AI systems, with large spreadsheets of art styles being passed among prospective AI artists. The more people know about art history and art styles in general, the more powerful these systems become. And people who respect art might be more willing to refrain from using AI in ways that ape the style of living, working artists. So a deeper understanding of art and its history can result not just in better images but also, hopefully, in more responsible ones.
A lot of work is time-consuming by design. In a world in which the AI gives an instant, pretty good, near universally accessible shortcut, we’ll soon face a crisis of meaning in creative work of all kinds.
We start to create documents mostly with AI that get sent to AI-powered inboxes, where the recipients respond primarily with AI.
This kind of meaningless task, what organizational theorists have called mere ceremony, has always been with us. But AI will make a lot of previously useful tasks meaningless. It will also remove the facade that previously disguised meaningless tasks.
Only 36 job categories out of 1,016 had no overlap with AI. Those few jobs included dancers and athletes, as well as pile driver operators, roofers, and motorcycle mechanics (though I spoke to a roofer, and they were planning on using AI to help with marketing and customer service, so maybe 35 jobs).
So, regardless of its nature, your job is likely to overlap with AI in the near future. That doesn’t mean your job will be replaced. To understand why, we need to consider jobs more carefully, viewing them from multiple levels. Jobs are composed of bundles of tasks. Jobs fit into larger systems. Without considering systems and tasks, we can’t really understand the impact of AI on jobs. Take my role as a business school professor. As the 22nd most overlapping of 1,016 jobs, I am a little concerned. But my job isn’t just a single, indivisible entity. Instead, it comprises a variety of tasks:
...more
However, this isn’t the end of the story. The systems within which we operate play a crucial role in shaping our jobs as well. As a business school professor, an obvious system is tenure, meaning that I cannot be easily replaced, even if my job were outsourced to AI. But more subtle are the many other systems at a university. Let’s say an AI could deliver a lecture better than I can. Would students be willing to outsource their learning to AI? Would our classroom technology be able to accommodate AI teaching? Would the deans of the university feel comfortable using AI in this way? Would the
...more
But a more careful look at the data revealed something both more impressive and somewhat worrying. Though the consultants were expected to use AI to help them with their tasks, the AI seemed to be doing much of the work. Most experiment participants were simply pasting in the questions they were asked, and getting very good answers. The same thing happened in the writing experiment done by economists Shakked Noy and Whitney Zhang from MIT, which we discussed in chapter 5—most participants didn’t even bother editing the AI’s output once it was created for them. It is a problem I see repeatedly
...more
Dell’Acqua developed a mathematical model to explain the trade-off between AI quality and human effort. When the AI is very good, humans have no reason to work hard and pay attention. They let the AI take over instead of using it as a tool, which can hurt human learning, skill development, and productivity. He called this “falling asleep at the wheel.”
The future of understanding how AI impacts work involves understanding how human interaction with AI changes, depending on where tasks are placed on this frontier and how the frontier will change. That takes time and experience, which is why it is important to stick with the principle of inviting AI to everything, letting us learn the shape of the Jagged Frontier and how it maps onto the unique complex of tasks that comprise our individual jobs.
remember the fourth principle: this is the worst AI you will ever use.
We have already seen that there is not a clear bright shining line of “human things” that the AI cannot do. It does a good job approximating empathy, creativity, and humanity. Trying to find things that AI can definitely not do because they are uniquely human may ultimately be challenging. But that doesn’t mean we want AI to do all these things. We may reserve Just Me Tasks for personal or ethical reasons, such as raising our children, making important decisions, or expressing our values.
I gave the AI the text of this chapter, up until this sentence, and asked it: Describe my writing style. It told me that my writing is a blend of academic rigor, personal insight, and practical advice, presented in a conversational, accessible manner.
Until AIs become very good at a range of Automated Tasks, the most valuable way to use AI at work is to become a Centaur or Cyborg.
Centaur work has a clear line between person and machine, like the clear line between the human torso and horse body of the mythical centaur. It depends on a strategic division of labor, switching between AI and human tasks, allocating responsibilities based on the strengths and capabilities of each entity.
Cyborgs blend machine and person, integrating the two deeply. Cyborgs don’t just delegate tasks; they intertwine their efforts with AI, moving back and forth over the Jagged Frontier. Bits of tasks get handed to the AI, such as initiating a sentence for the AI to complete, so that Cyborgs find themselves working in tandem with the AI.
I am only human, and in writing this book, I often found myself stuck. In previous books, that could mean a single sentence or paragraph would block hours of writing, as I used my frustration as an excuse to take a break and walk away until inspiration struck. With AI, that was no longer a problem. I would become a Cyborg and tell the AI: I am stuck on a paragraph in a section of a book about how AI can help get you unstuck. Can you help me rewrite the paragraph and finish it by giving me 10 options for the entire paragraph in various professional styles? Make the styles and approaches
...more
reading papers was often a Centaur task, one in which I knew the AI exceeded my capabilities in summarizing, while I exceeded it in understanding.
One small study of undergraduates found that 66 percent of men and 25 percent of women choose to painfully shock themselves rather than sit quietly with nothing to do for 15 minutes. Boredom doesn’t just lead us to hurt ourselves; 18 percent of bored people killed worms when given a chance (only 2 percent of non-bored people did). Bored parents and soldiers both act more sadistically. Boredom is not just boring; it is dangerous in its own way.
Stock photography, a $3 billion per year market, is likely to largely disappear as AIs, ironically trained on these very images, can easily produce customized images. Or consider the $110 billion a year call-center industry, which will reckon with the impact of fine-tuned AIs handling ever more tasks that were once done by humans, acting like a phone tree service that actually works. At the same time, entirely new industries may appear, like those servicing and deploying AI systems.
By the 1920s, 15 percent of all American women had worked as operators, and AT&T was the largest employer in the United States. AT&T decided to remove the old-school telephone operators and replace them with much cheaper direct dialing. Operator jobs dropped rapidly by 50 to 80 percent. As might be expected, the job market overall adjusted quickly, as young women found other roles, like secretarial positions, that offered similar or better pay. But the women with the most experience as operators took a larger hit to their long-term earnings, as their tenure in a now extinct job did not
...more
