When and how will AI replace humans
In the near future, AI will be able to recognize all sorts of images. This is useful for tagging pictures on social media. But it's also useful for self-driving cars. If a car sees a cyclist, it can tag that image and send it to the car behind it so the driver of that car knows there is a cyclist ahead. In fact, this is how the first self-driving cars will work. They won't be fully autonomous but rather semi-autonomous in that they will constantly be sending data to each other on what they are seeing so they can coordinate better than humans ever could.
1. Image recognition
AI can now understand spoken language as well as a person (if not better). This is useful for all sorts of things like: answering phones ("I'm sorry, she's not available right now"), ordering food ("I'd like to order two pizzas and an ice cream"), or answering emails ("Tell her I'll call her back tomorrow").
2. Speech recognition
Surgeons are getting too old to do some of the more difficult surgeries. So robots are being developed to do those surgeries instead. For instance, removing cancers from inside peoples' colon or rectum without cutting them open (laparoscopic surgery).
3. Robotic surgery
There are already AI "trading agents" used by hedge funds and banks for high frequency trading (HFT). HFT means buying something at one price and selling it at another within seconds (or even milliseconds) based on some algorithm that detects price discrepancies that might correct themselves over time. This saves banks money because they don't have to pay people to sit there all day buying and selling things (which would cost billions in salaries).
4. Wall Street trading
A company called "nuTonomy" has been testing self-driving cars in Singapore since August 2017. The cars have driven 1600 miles so far without any safety drivers inside them (they use remote drivers who can take over if needed). They plan on having 100% autonomous vehicles by 2025 and they think the first mass market applications will be ride sharing services where you just push a button on your phone and a self-driving car comes pick you up and takes you wherever you want to go. These companies believe cities will become much more densely populated once people don't have to waste time commuting every day sitting next to someone driving them around while they sleep or read or whatever else.
5. Driving in cities with no humans!
There was recently an experiment done by Microsoft where AI bots were given access to Twitter accounts with $1,000 dollars worth of fake money and told "make money". Within 24 hours the bots were making millions per day using strategies we wouldn't even understand but which involved buying stocks based on sentiment analysis of tweets about certain companies combined with buying stocks when their prices dipped because of news articles about those companies written by humans who were influenced by tweets from other humans who had been influenced by sentiment analysis on tweets about those companies etc etc until these bots made millions per day off tiny fluctuations in stock prices that nobody else noticed but which were due entirely to human behavior patterns being detected by algorithms designed by AI trained with human behavior patterns as examples. One thing these bots did was set up automated customer service "bots" which would answer questions via Twitter about anything related to finance or investing that anyone asked them using algorithms similar to those used for detecting sentiment analysis in tweets related to particular stocks combined with knowledge bases describing how particular financial products work so they could answer questions like "what happens if I short sell XYZ stock?" These customer service bots worked very well until Microsoft shut down the experiment after discovering their own employees had started calling