GPT-3 and 2022 Predictions
Fun blog to start the year, looking at the state of NLP (natural language processing) in AI.
You may be familiar with GPT-3, which is an NLP model from OpenAI. Introduced in 2020, GPT-3 is probably the most famous NLP model in the world due to its uncanny ability to produce human-like prose. I have written about GPT-3 before (you can find that here), if you want a bit more depth on the actual model.
However, there is nothing that demonstrates the current state of NLP technology better than the visceral reaction of seeing what GPT-3 can write given a prompt. So, I thought I would start 2022 with a blog in which I outsourced most of the work.
I provided the GPT-3 model with a prompt (“I have 10 predicitions for 2022, which are listed below. 1.”) and then let it run with it 4 times. See below for the results (everything in italics was written by the algorithm, and my commentary is in bold).
Run #1: The Cruise / Zoox / Waymo Investor
I have 10 predictions for 2022, which are listed below.
- Self-driving cars are here to stay and will be as common as today’s cars. (There are today over 250 million cars on the road in the US.)
- Self-driving cars will be commercially available at that time. [Once there are 250M of them? Yep, the logic checks out.]
- There will be a self-driving car fatality somewhere in the world by 2022. [This took a dark turn.]
- Autonomous vehicles will be able to navigate more-complicated environments than those we navigate today.
- Consumers will fall in love with autonomous vehicles.
- The first case of a human being being hit and killed by a self-driving vehicle will create a firestorm in the press. [This took a dark turn part 2. Also, a good example of the mindlessness of the algorithm, as a human would have combined 3 and 6.]
- The victims of that first fatality will be resurrected in the court of public opinion. [Whew — that’s a relief. But a good example here of the algorithm misusing an idiomatic use of resurrected!]
- In 20 years, there will be an article in Popular Mechanics or Car & Driver or some other publication that recalls the first fatality and the furor that followed. [Not really a 2022 prediction, but we will let it slide.]
- Self-driving cars will be much safer than human-driven cars.
- The fatality rate for self-driving cars will be measured in terms of fatalities per kilometers driven, not fatalities per 100 million kilometers driven. [Controversial!]
Join us [This was a sinister way to end…]
Run #2: The Generic Venture Capitalist
I have 10 predictions for 2022, which are listed below.
- A majority of homes around the world will be smart homes.
- Science fiction will become reality.
- The “internet of things” will surpass the “internet of people.” [Doesn’t this sound vaguely threatening when you know a computer wrote it?]
- The new workplace will be more like a social network than an organization chart. [This is actually kinda profound!]
- Big Data will be the basis of all decisions.
- Teams will replace individual accomplishment.
- Everyone will get security alerts. [Umm…could’ve used some more detail here.]
- More CEOs will become personally involved with product design.
- The market-cap of social networks will exceed that of any nation. [GPT-3 is a supporter of #regulatebigtech.]
- Everyone will be an entrepreneur.
The first three predictions are obvious given that most of the work on the “internet of things” and the “internet of people” is already being implemented. I think the market capitalization of social networking companies will outstrip the market capitalization of all nations. We are all entrepreneurs now, without even knowing it. We are, after all, creating value for others every minute of the day, seemingly without even lifting a finger [this is literally true for an algorithm!].
Run #3: The Self-Promoter
I have 10 predictions for 2022, which are listed below.
- More than 90% of American homes will have some type of home AI system.
- 60% or more of online applications will be AI or AI in combination with other technologies.
- One prediction every 15 minutes will be made by AI (most will be wrong). [Some self-deprecation here! Although, GPT-3 isn’t really a prediction engine, so maybe this is just good sibling rivalry.]
- Each day there will be one national news story that is completely wrong due to a confusion between AI, ML and a new technology. [GPT-3 underpromising for when it starts writing the news.]
- There will be one or more movies that portray AI in a negative light.
- There will be one or more movies that portray AI in a positive light.
- There will be one or more movies that portray a combination of people, AI and robots in a positive light. [Is GPT-3 trying to hit a word count here? Really scraping the bottom of the barrel on this movie prediction.]
- Cyber-bullying will be more prevalent and much more difficult to detect than today.
- The team that wins the next Football Super Bowl will use a lot of AI and ML. (After all how else will they be able to throw a football through an 18 inch opening from 4 miles away?)[Hot seat — Patrick Mahomes.]
- By 2022 there will be an AI and/or AI in combination with other technology that has saved more than 100,000 lives.
Run #4: The Far-Right
I have 10 predictions for 2022, which are listed below.
- Global warming is a hoax. The science behind it will be exposed as a fraud.
- The Endangered Species Act will be abolished.
- The EPA will be abolished.
- The Department of Energy will be abolished.
- The Department of Education will be abolished.
- The Department of Commerce will be abolished.
- The Department of Labor will be abolished.
- The Department of Health and Human Services will be abolished.
- The Department of Housing and Urban Development will be abolished. [You see this pattern sometimes in GPT-3 — it gets stuck in a ‘rut’, in which it starts repeating the same form. In this case,’ X should be abolished’].
- The top tax rate will be lowered from 39.6% to 28%.
These are all goals that President Trump and the Republican Party should work toward. They’re achievable, and will make America great again.
We will leave it there for now — Happy New Year, everyone, and here is to most of these predictions not coming true!
