I recently read the 2024 book “Welcome to AI: A Human Guide to Artificial Intelligence” by David L. Shrier and here’s a quick summary of quotes and important points.
In a similar vein to “The Coming Wave” by Mustafa Suleyman (which I talked about recently) but with a more political and societal focus, instead of a technological one, it’s an interesting book that I can recommend. It tries to provide an answer to an uncomfortable question: what will happen to jobs, democracies, governments, and education, in a world where AI exists and becomes more and more capable by the minute?
The book starts with the enumeration of three watershed moments in AI history:
- The release of TensorFlow by Google in 2015.
- The publication of the BERT paper by Google in 2018.
- Finally, the release of ChatGPT by OpenAI in 2022.
I found interesting the stress on TensorFlow as one of the catalysts of this new era in computing, a point the author repeats across the book.
Chapter 1: Rise Of The Bots And AI Hallucinations
Let’s start with a simple fact:
ChatGPT has been the most successful new consumer technology launch in history.
However,
This reveals an important and fundamental characteristic of artificial intelligence: its lack of reliability.
Chapter 2: Definining AI
Ouch:
The reason most chatbots today seem dumb is because they are.
TensorFlow is fundamental in the big scheme of things:
TensorFlow democratized access to powerful, fast machine learning capabilities.
Remember CAPTCHAs? Turns out that we’ve been effectively teaching AIs for decades now. The scale, however, is much bigger than I would have thought!
As of 2019, over 4.5 million websites had embedded reCAPTCHA, delivering more than 100 person-years of labor every single day, labeling and classifying image data.
(…)
The value of an individual image classification or annotation can be $0.03, so you could estimate $21 billion or more in labor arbitrage that Google has extracted by having people train AI.
Chapter 3: Evolution Of AI Jobs Displacement
Good perspective. I personally know friends in Argentina who have literally lost all income from their design or copywriting personal business during 2023. This is happening.
Conversely, the developing world, already burdened with issues like population growth, insufficient food and water supplies, and inadequate education, is being more severely disrupted by automation than the developed world.
But this is coming to the developed countries as well, even to lawyers:
Billable hours, the holiest of holies of the legal profession, will evaporate like water on a hot griddle.
And even to… machine learning engineers:
Google, of course, is at the cutting edge of this field, creating a kind of AI called AutoML, that can program itself without human intervention.
Chapter 4: Reskilling and Developing Cognitive Flexibility
Schools and universities are definitely not ready for AI.
Our current educational system does a good job of training the creativity and mental agility out of us by puhishing those who color outside the lines and rewarding conformity.
Yes, I agree. There are better ways:
One of the best ways to learn is through peer education. Medical education, for example, runs on the principle of watch one, do one, teach one.
(…)
One of the benefits of peer learning is that if you are forced to explain a subject to someone else, you tend to understand it better yourself.
The key lies in the emotional part of our brains, and our interconnections to others:
Instead, the teams that performed the best, that had the highest collective intelligence, were those with an exceptionally high emotional IQ or EQ.
(This might be a problem in Switzerland, if you ask me.)
Chapter 5: Future Proofing
This is by far the most important chapter of the book. What current professions and jobs are at stake in the age of AI? Turns out, a lot of them.
The accounting professions, once considered among the safest career paths, actuarial students and even pre-law, as we reviewed earlier, all are meaningfully at risk.
Ironically enough,
In a world where AIs assume repetitious, analytical tasks, soft skills assume greater importance. This reverses an educational trend of the past thirty or so years favoring preprofessional tracks like finance or business or computer programming, increasingly steering students into semi-vocational training for perceived safe jobs.
Will AI help us recover our more human traits?
But the quote of the book that sums it all, is this one:
Where there are people, there is labor. Where there is labor, there is cost, and with cost, the inexorable forces of capitalism and the public markets will drive toward creating market efficiencies, that is, reducing labor.
Yup.
If you want to win at work in the age of AI, you need to find ways to make AI work for you, rather than against you, augmenting with your capabilities, rather than seeking to replicate them.
Chapter 6: What Can Human+AI Systems Do?
The author proposes the “augmentation” of workers through AI, and explains how this could happen.
To truly bring together the fusion of human and artificial intelligence systems, we need to take the relationship a step further.
(…)
The first step is to have a heightened awareness of what you do at work and how you interact with technology.
And a funny observation:
Legions of disappointed Game of Thrones fans, myself included, would gladly re-create the final two seasons to provide for a more satisfying denouement.
Chapter 7: Building Human+AI Systems
More seriously, how do you kickstart the augmentation? This chapter deals with the new concept of “hybrid teams” mixing humans and AI to fulfill business tasks. Arguably, not a future, but a reality today.
Prompt engineering, the art of how you construct a conversation with generative AI, is a new essential work skill.
Chapter 8: Alternate Futures
At present, there is no global standard for the application of trusted or ethical AI.
Ethics, turns out, is at the core of our issues with AI:
The lack of ethics instruction in artificial intelligence programmers has already shown itself to be harmful to humanity.
Yes, I agree, again.
Governments have a big role to play in this new era:
The role of government is all too often diminished by free-market advocates and by the media, but shaping the trajectory of AI in society is the type of large-scale public service for the common good that government is ideally constituted for.
Chapter 9: The Urgent Need for Policy
Continuing the previous idea:
Government has been the primary funder of the AI revolution, after all. Shouldn’t it be responsible for what it has helped create?
Some pundits might reject the possibility of this idea; but numbers don’t lie:
For the 2022 fiscal year, 67 percent of MIT’s core research funding of $783 million was US federal government funding–as well as another $1.1 billion at MIT Lincoln Lab, a government-only affiliated lab.
Governments all over the world are taking note:
More than 800 AI policy initiatives have been actioned in over sixty countries, most of them since 2016.
But…
If the policymakers lack fundamental understandings of how the systems work, how can they be expected to regulate them?
The author proceeds to enumerate some concrete steps for governments to follow to encourage positive AI developments, which I won’t enumerate here. Read the book!
If there’s one idea I want you to take aways from this book, it’s the message of hope coupled with responsibility.