Even programmers are starting to fear that ChatGPT will replace them. How real is the threat?
Everyone has already realized that the painters and artists fear for the future of their work because of artificial intelligence, but in the meantime - quietly - the programmers and developers are also starting to develop anxieties. In the professional groups on Facebook, you can find plenty of posts written with wide eyes describing how ChatGPT or other artificial intelligence, generate code and solve development challenges without thinking twice, or at all.
Unlike the artists, the general tone of computer science people is that of excitement and enthusiasm. This is not surprising: they have learned to master the tools of artificial intelligence. At the same time, even with them you can begin to hear implicit concerns. If artificial intelligence knows how to write code, will there really be a need for programmers in the future? Is even a computer science degree a guarantee of a good job?
A researcher named Michael Webb realized already three years ago that computer science may be much more vulnerable to appeal than it seems at first glance. Webb conducted a comprehensive analysis of more than 16,000 patents in the field of artificial intelligence from recent years, and compared the abilities described there to the abilities required in the various professions in the US. He recognized that precisely the white-collar professions - which are considered more profitable and durable than the blue-collar professions - are very exposed to change[1].
Until Webb, the common assumption was that it was blue-collar workers - such as welders, cooks and taxi drivers - who would be in trouble in the coming decade. This is probably still true, but Webb has added another layer of complexity to the picture. He found that artificial intelligence is rapidly improving at performing non-routine tasks in the digital world. She is able, for example, to write code, develop scenarios, express positions and opinions in writing, generate attractive slogans, develop websites, and much more. And although it is far from perfect in each of these, the trend is already clear. Artificial intelligence is improving every week and every month, and there is no reason to assume that we will reach the limit of its capabilities in the coming years... or at all.
I remembered all this background when one of my friends sent me a WhatsApp message this morning. To protect his privacy we will call him L. He manages a software division at one of the largest and oldest hi-tech companies in the world, and as part of his position he is responsible for sixty employees - almost all of them engineers, researchers and developers. In other words, if you want to find a job in high-tech, you will probably have to go through someone like L in the early stages of your career.
So you should know that in the last few days, L's manager asked him to check how many people he still needs in his department. And as he wrote to me -
"My manager asked me to see if I could replace junior programmers at chatGPT."
Of course I asked him to expand.
It is important to clarify that L. did not conduct a thorough or in-depth examination, and he is only sharing some simple tests he did and his thoughts. Still, this is an interesting angle from a person who has to decide whether to recruit more programmers to the team, or whether artificial intelligence can fill their place.
So what does he think?
"I asked the bot to write me a code with specific parameters I need." L. shared. "I indicated in which language and with which libraries I wanted to work. Wrote me great code. Not the height of elegance, but better than engineers at the beginning."
L. tried two tools - ChatGPT and Github CoPilot and was impressed by the capabilities of both.
"They both did the job quite impressively." He said in a conversation with me. "The way I worked with them is to ask for code that does something in the language I need and see if the bot is in the right direction. As an example, I asked the bot to write me code to manage a set of tests. The result was largely in the direction I wanted."
The place where L's eyes began to open - and he realized the true potential of these tools - was when he started having discussions with the artificial intelligence to improve the code.
"It's not all or nothing." He explained to me enthusiastically. “I got code in the right direction, then asked the bot to make changes to get to the final product. For example, I asked him to add an API with the Flask library. He did it really easily. I could ask him for changes like I ask them from the engineers I employ."
He paused for a moment, then added - "Only unlike working with a junior engineer, I didn't have to teach the bot anything." With junior engineers I spend a lot of time being a mentor. I didn't have to here."
L. is not innocent: he understands that artificial intelligence will not replace his engineers tomorrow morning. It still has plenty of limitations.
"We wanted to see how artificial intelligence can help us in development, especially in areas where the code is not innovative or too complex." said. "Most of what we write on a daily basis is code that is fairly standard in the industry, and writing it is more a matter of sitting down to write and less of a matter of deep thinking."
Even in these cases, L noticed that the AI didn't do a great job. Just good enough.
"I caught some small problems that someone who doesn't write code wouldn't see." said. "So if I didn't know anything about programming, I could get a slightly inferior product, but it works. You need to know how to read the code he gives you, and be capable of analytical thinking to see if it solves your problem or if there are holes in the logic."
Artificial intelligence is not perfect, therefore, but L did not expect it to be either. He just wanted to figure out if he and his people could rely on her to do the simpler, more menial tasks of development for them. In that respect, she definitely passed the test. Those "simple and menial tasks", of course, are the ones he had to hire junior programmers for. And this is just the beginning.
"Our big goal here was to see how we integrate with this technology instead of fighting it." He explained candidly. "I understand that urge. I myself am not really sure what my field will look like in 5 years. We embrace technology with open arms right now to stay relevant. Those who fight it and don't understand that the paradigm of work is changing here, are going to be left behind."
I stopped him at this point, and asked to present to him the "big winners" model. This unofficial model is based on the ideas of economists who study the future of work, such as Taylor Cowan ("Average is Over")[2] Varic Brynjolfsson and Andrew McAfee (“The Second Machine Age”)[3]. According to the model, there are several ways to become a "big winner" in the future - one whose services companies and people in general are interested in and need.
The first way is to acquire a high level of skill, of the type that usually characterizes people with a master's or third degree in their field. Highly skilled graphic artists, for example, know more than 'just' how to draw. They know the many theories that describe aesthetics, understand the different styles for painting and know how to adapt them to the task at hand. Thanks to this extensive knowledge, they are able to formulate a vision for the desired work - and explain to the artificial intelligence what it should do. When she produces the work for them, they can go over it and identify all the small problems, or the places where the product does not match the vision. One highly skilled person can in this way replace an entire department of developers, graphic artists, or futurists.
The second way to success goes through the soft skills, and specifically - motivation, curiosity and the ability to learn. The big winner of the future does not have to have a PhD in computer science. He can also be an ordinary person with a lot of motivation and with the ability to acquire new skills at a basic level. Such a person - Jack of all trades, as he is called in English - will be able to use any new technology to upgrade his current way of working. A taxi driver could easily create a bot that would answer customers on the phone with his name - and even his style. A 16-year-old girl will be able to develop, in a week's work, an application that would previously have required an entire development team. And as soon as she sees that the application is successful, she will also build a start-up around it and recruit the real experts with the money she earned.
The third way to success is in developing the ability to think strategically, analytically and critically. We will all be working with artificial intelligence in the near future. Some of us will actually use them, and the rest will have to deal with the products they produce. When artificial intelligence can produce a huge abundance of products for us, the user must think ahead and understand what exactly he wants and needs. This is strategic thinking. He should be able to understand when the artificial intelligence is wrong and misleading, and when it approaches the truth - but it still needs guidance. It is not surprising to find that the most required skill in the market today is "analytical thinking and innovation", followed by "learning strategies and active learning", "complex problem solving" and "critical thinking and analysis"[4].
What did L think - the manager who will decide whether to hire you to work at Microsoft, Facebook or IBM?
"Everything you said is true, except for the first clause." fix. "You said that artificial intelligence will replace engineers, and whoever will use it will be able to take the place of many other people. But the bot still requires a lot of tinkering. Human engineers can exercise more discretion. You give them work direction and free them to create things you never dreamed of. If the bot was able to understand the next logical step or give some solutions according to the next steps it would give a complete alternative to a human engineer. I see this bot right now as complementary to some actions of engineers that can significantly reduce their workload. Let's say a bot that reviews the code before adding it to the main code base. It takes a lot of human time right now and is full of misses.”
Well and good, I said. Even two out of three is considered a success. But what will happen when the artificial intelligence can also understand what the "next logical step" is, or solve problems according to patterns that are fed to it, or that it will recognize by itself?
He shrugged.
"I guess it will happen within a decade." said.
Is he right? Maybe, but I'm more optimistic in terms of schedules. Already today it is possible to see the "hyper-automation" of complex problems: that is, the decomposition of a complicated process into its components, when each part of it is realized by a different artificial intelligence. Such hyper-automation, when it is based on the increasing power of the new artificial intelligences, will also be able to occupy the workplaces of the high-tech people who do not benefit from an excess of skill, motivation or strategic and critical thinking.
Should the hi-techists worry? It doesn't seem that L is excited.
"I don't worry about people in high-tech." fix. "People who work in our field are used to changes. It is part of our life. I believe they will be able to adapt."
Maybe he's right, but that's exactly the point: we all need to understand that we need to adapt to the new reality. We cannot rest on our laurels or assume that the world will continue to operate as it has been until now. We must perfect ourselves, acquire new skills and ways of thinking, adopt the technologies into our bosom - or develop a clear strategy to deal with them. Whoever chooses to close his eyes even for a moment during the race, will stumble and stay far behind.
We are not ants, who cannot change their own programming. We are human beings: flexible and adaptable creatures who gather and build skills throughout life and change the way we define ourselves every year. Artificial intelligence is going to disrupt our lives, and that's okay. It is allowed to be challenged. Then, you have to embrace the change with both hands and use it to grow and move forward.
Artificial intelligence will be the great destabilizer of the current decade, and it will also be the springboard for many to better and more interesting jobs. We can all be part of the big winners - but we need to start acting for it now. To acquire a high level skill, develop and maintain curiosity and motivation and think strategically and critically.
And if you figure out how to do all this, please tell me too.
[1] https://www.brookings.edu/research/what-jobs-are-affected-by-ai-better-paid-better-educated-workers-face-the-most-exposure/
[2] https://www.amazon.com/Average-Over-Powering-America-Stagnation/dp/0525953736
[3] https://www.amazon.com/-/he/dp/B00HFX8JMQ/
[4] https://www.weforum.org/agenda/2020/10/top-10-work-skills-of-tomorrow-how-long-it-takes-to-learn-them/
