Machine learning is a process in which an AI can become better at performing a certain task by being given hundreds to thousands of examples of the task being performed correctly, until the machine “learns” how to amend itself toward ideal functioning. A simple example of machine learning is that after being given thousands of samples of drawings, a machine can learn to accurately state what’s being depicted in each drawing. After an incorrect guess, the machine is given the correct answer and changes how it processes the data accordingly.
But then, that isn’t all machine learning is good for. Read on for some new ways this technology is being applied:
Finding People in a Crowd
When connected to a database of faces and a CCTV camera, there are now some machine-learning-based AIs that can accurately pick someone out of a crowd while they’re moving. The machine learns how to compare pictures of people from one angle to the many different angles they may present themselves while in public, and then point them out to those monitoring the cameras. This has an application in debt-collection, crime-solving, or even pre-approving specific individuals to enter restricted areas without the need for a keycard. Some worry this technology will be abused by government powers, but this effect is yet to be seen.
Automating Trades
Of course, the financial industry wishes to capitalize on the opportunities presented through this technology, as a sufficiently “educated” AI can operate at a much higher capacity than the average human. An example is the site bitcoin-revolution.co which uses an automated system to automatically conduct trades that will result in a profit. Money is put into the system by users, and a small percentage is taken by the website for giving users the privilege of access to their bot. Over time, the bot will be able to make better and better choices, resulting in more profit for everyone involved.
Read Also: Top 10 Trading Tips for Beginners
Personalized Health Care
One of the most novel applications for machine learning is in the realm of health care. It’s where robots can be educated to quickly suggest the best treatment plan for any particular patient in a manner that’s much more complementary to their current situation, instead of using general rules of thumb. As the robot deals with more patients with the same condition and weighs their outcomes versus the treatment options used, the robot will tend toward plans that have higher success rates for an individual patient. These systems might not be impressive at launch, but in a decade or so, we wouldn’t consider it strange for a doctor to defer to an AI for their next move.
Conclusion
Previously, nearly all improvements to an AI’s functioning had to be done manually and in a piecemeal fashion as new issues arise. These new AIs can change their coding themselves in real-time to better deal with situations, instead of needing a constant helping hand. Once the right safeguards have been put in place to avoid negative situations, there are essentially only positive outcomes to a properly coded AI. Who knows, maybe the next article you read will have been written by a bot, and there won’t even be a way for you to tell.
