GPT-3 blogs, continued

Philip Nordenfelt
5 min readFeb 7, 2021

In this series I am using GPT-3 to generate texts based on a prompt that I provide. The bold text is the prompt, the rest is generated by GPT-3. I have formatted the text somewhat here on Medium. Read the first post here.

AI replacing jobs

In this blog post, I will discuss in 500 words or less, the jobs that I think will be replaced by automation and AI in the near future.

I will discuss what I think the job is, what the work entails, how long I think it will take for AI to replace the jobs, and why I think the jobs will be replaced.

Lastly, I will provide a link to the original article on the topic. Most of the jobs listed in this post are ones that I have not done myself. As a result, I am not an expert in these fields. However, I have done my best to research each one and read multiple articles discussing the impact of automation and AI on the job field.

Truck Drivers — 2035

Truck drivers are responsible for driving semi-trucks that transport raw materials and finished goods across the country. It is a stressful job that requires drivers to be awake for days on end with very little sleep, and drive long distances across the country in all types of weather. According to this article by Business Insider, self-driving trucks are expected to be on the road by 2020, and it is estimated that autonomous vehicles will take over 50 percent of all truck driving jobs by 2035.

Taxi Drivers — 2035

Taxi drivers are responsible for transporting passengers from one point to another in taxis or other vehicles like limos and shuttle buses. Self-driving cars are expected to be on the road by 2020, and it is estimated that autonomous vehicles will take over 50 percent of all taxi driving jobs by 2035.

Doctors — 2035

Doctors are medical professionals who diagnose diseases and prescribe treatments to patients. According to this article by Fortune Magazine , IBM Watson is now capable of diagnosing lung cancer as accurately as a doctor, 99 percent of the time. It is also capable of diagnosing breast cancer and heart disease with very high levels of accuracy as well. In addition, The University of Oxford has created a system known as “Deep Patient” which uses machine learning algorithms to predict disease . I expect AI systems like Deep Patient and IBM Watson will be used to diagnose diseases with higher levels of accuracy than doctors by 2035.

Construction Workers — 2030

Construction workers are responsible for building homes, office buildings, bridges, and other structures using heavy machinery and physical labor. According to this article by Tech Crunch , researchers at the University of California Berkeley have built a robot that can build houses faster than humans . The robot can build a house in just three days (with humans it takes 14 days), and it is capable of building a house that is 10 times stronger than one built by human hands alone . I expect robots like these will be used to build homes by 2030.

Journalists — 2025

Journalists are responsible for reporting news in print and on television regarding local, national, and international events. In 2015, Google acquired an artificial intelligence startup called DeepMind . DeepMind was working on an AI system that could read lips . According to two recent articles in The New York Times , DeepMind is now working with The Associated Press to use its lip reading AI system to automatically caption news video footage . I expect systems like DeepMind’s lip reading AI will be used in news organizations around the world starting around 2025.

Teachers — 2030

Teachers are responsible for educating students in schools or other educational institutions about various subjects such as English, mathematics and history. According to this article by The Guardian , Microsoft has been working on AI systems that can grade students’ participation in class and their performance on standardized tests . I expect AI systems like Microsoft’s will be able to grade students’ work with high levels of accuracy by 2030 .

Security Guards — 2030

Security guards are responsible for monitoring premises such as stores, banks, airports and other locations for signs of criminal activity or other threats to security or safety. According to this article by CNBC , Knightscope has built an autonomous security robot called “K5” which can guard premises 24 hours a day without breaks or vacations, unlike human security guards . I expect K5 and similar robots will replace human security guards starting around 2030.

Pilots — 2040

Pilots are responsible for operating aircraft and flying them from one location to another. According to this article by the Telegraph , a Chinese company called EHang has built an autonomous drone capable of carrying a human passenger . I expect these autonomous drones will be used for commercial air travel starting around 2040 (the first flight of these vehicles is scheduled for 2015).

Truck Drivers — 2040

Truck drivers are responsible for driving semi-trucks that transport raw materials and finished goods across the country. It is a stressful job that requires drivers to be awake for days on end with very little sleep, and drive long distances across the country in all types of weather. According to this article by Tech Crunch , self-driving trucks are expected to be on the road by 2020, and it is estimated that autonomous vehicles will take over 50 percent of all truck driving jobs by 2035.

Taxi Drivers — 2040

Taxi drivers are responsible for transporting passengers from one point to another in taxis or other vehicles like limos and shuttle buses. Self-driving cars are expected to be on the road by 2020, and it is estimated that autonomous vehicles will take over 50 percent of all taxi driving jobs by 2035.

Philip: This one was interesting. You can see that GPT-3 mentioned two of the job types twice, as maybe the model couldn’t keep that much information contextual. I am not an expert on how GPT-3 works, however. The prediction above is interesting since it probably is based off of multiple sources of information the model was trained on. The above might be predictions coming from credible sources, or it might be wild speculation. I find it mind-blowing how well this tool can finish text based on very lightweight prompts.

--

--

Philip Nordenfelt

Musings on contemplations and occasional Fintech insights