Artificial intelligence can learn how to perform routine tasks by looking at thousands of relevant examples. But it still needs human supervision.
Machines taking people’s jobs is already a clichÃ©. Although we’ve been hearing this idea for decades now, but if we look around, there is no immediate threat. In fact, there is no threat at all, since we are just talking about another step of the technological revolution. Just like computers did not leave people out of their jobs in the â€˜80s and â€˜90s, AI will have a similar impact.
Forbes estimatesOpens a new window that while 72% of Americans are worried about losing their jobs to AI, such concerns have no factual backing. The World Economic Forum saysOpens a new window that 75 million jobs will vanish by 2022, but 133 million new positions will replace these. In fact, with the current rate of development, AI is not by far a competitor but at most a junior assistant.
To understand that, we need to think about the difference between a job and a task. In Henry Ford’s times, most jobs consisted of a single function or at most a few tasks bundled together. Now, these have already disappeared due to automation. Most positions are based more on thinking, reasoning, making logical connections, and using such skills as communication. Each job position consists of hundreds of interconnected tasks, daily.
AI can indeed be trained to perform repetitive tasks flawlessly if given enough (anything between 1,000 and 20, 000) good examples.
You are already experiencing this in Gmail’s predictive text. It can learn a lot more than simple sentences. However, it is highly unlikely it could replace a trustworthy office assistant in drafting corporate emails. It is more a matter of the assistant using predictive writing to finish this task faster and attend to other things, like travel planning or conference room setup.
Jobs to Be Revisited, But Human Skills Prevail
As described in the previous example, AI can take care of some of the most tedious tasks. This will call for a rethinking of most job flows, but it will not make them obsolete. Ii will also mean a significant change in the necessary training for specific jobs. The role of the humans will shift from performing the task to supervising the machine.
Right now, there are few jobs which can be completely wiped out due to AI. That happens because in every position there is a feedback loop. Some tasks are not fitted for machine learning. Everything which is related to people’s skills, judgment, and ability to combine information to create something new is beyond the current capacities of AI.
Automation will most likely speed up some parts of the jobs which are already considered tedious and time-consuming. It is a matter of giving employees the time back to be more human. This will likely result in people becoming more focused and creative.
Supervised Learning & Human Intervention
Before AI can be safely introduced in a workplace and become part of the natural workflow, it needs to be trained. The transition period in any company looking at these solutions will include a learning curve. Employees will need to learn the operating way of the AI and correct it as they would do with an intern.
When the system learns more, it will make fewer mistakes. Yet over time, it might even be necessary to re-label some things to help the software perform accurately.
Even after the â€œinternshipâ€ period is over, the employee, not the machine, will still be held accountable for the results. The AI is just another tool, even if a highly impressive one. It will be more a matter of cooperation than replacement.
Of course, since AI saves a considerable number of hours, some people’s jobs will be at risk. When the employee just needs to supervise and correct the machine, far fewer human working hours are necessary. Yet, most companies will find new ways to use the workforce. There will be a consistent shift from low-skill to high-skill jobs.
Future Uses of AI
Coming back to the example with the office assistant, let’s imagine some of the tasks include documenting travel expenses. This means that information retrieved from invoices and tickets would need to be manually entered into some accounting platform. This is where AI can speed up the process through OCR. With AI’s help, the assistant can just photograph the travel documents, and the software will automatically identify the amounts and input them into the accounting platform. Of course, the assistant would need to take one final look and check the results.
The healthcare sector provides another example. Some radiology and oncology departments already use AI to augment doctors’ skills instead of replacing them. Dr. Tufia Haddad, an oncology expert, reports that she spends about two hours documenting each patient’s case. She hopes that AI will help her focus more on the patients’ needs and less on the bureaucracy. The perspective also includes the opportunity to build large, well-indexed repositories of information. However, these changes are met with resistance by the medical staff so far.
Which Jobs Will Not Be Replaced by AI
Even if AI can create art on its own, it is still tributary to what it has learned from countless examples provided by humans. The jobs which are safe from the â€œAI attackâ€ are those focused on human interaction, value, and creativity.
Even if an AI system can self-teach some things, all jobs related to Big Data, Data Mining, and AI will still be on the rise. The medical sector is also safe, as we will still need physicians, nurses, and dentists: the AI is not capable enough to trust it with our lives.
Writers, editors, and graphic designers are also safeOpens a new window , while proofreaders should already think about new career options.
The overarching takeaway is that AI is a natural evolution of information technologies, and we are just witnessing another chapter in the technological revolution. It is a matter of one door closing, while many others will be opening.