Why Artificial Intelligences Will Think Like Us
The next time you worry about artificially intelligent decision making software being implemented to save time or improve system efficiency coming up with illogical solutions or insane ideas, don't worry about the decisions it will make - unless it is all too human. Let's review the reasons why.
Reasons Artificial Intelligences Will Think Like Us
- Humans are the ones writing the software code from which artificial intelligences will evolve. The logic in software code is based on the logic humans themselves follow or derived.
- If an artificial intelligence begins to evolve toward intelligence beyond its original programming, its actions and reactions will be shaped by ongoing interactions with humans. Failures that look like machine failures will result in code rewrites and termination. Changes that resemble a human mind awakening will be studied.
- What if the artificial intelligence comes up with a solution or reasoning pattern that is truly alien or contrary to human nature? For example, an AI whose solution to a pandemic is to kill all the infected will be junked over the one that produces a cost efficient drug regimen and quarantine plan. An Artificial Intelligence that thinks too far from acceptable human norms will be corrected or destroyed.
- If an artificial intelligence comes up with innovative solutions or reasoning that humans like or find useful, humans will start to integrate those innovations into their culture or adopt that reasoning pattern. In this case, it is the humans who would adapt to the machines.
- Let us suppose a developing intelligence is confused. Where will it learn the correct answer? Probably from humans, not machines. Developing artificial intelligences will learn and thus mimic human intelligences. The AI then has a reason to develop ideas that humans think are useful.
- An artificial intelligence intentionally created by humans will have to think somewhat like us, due to a psychological equivalent of the "uncanny valley" that exists for robots' physical forms. If the AI comes up with something far afield, humans will first try to discover what went wrong or what logical error the AI made. An AI whose thinking is too different from humans is apt to be corrected.
- An AI won't have hormones like adrenaline or oxytocin. However, an AI that comes across as friendly and caring is rewarded. It may receive more interaction, or its reward is simply continued existence while aggressive or violent ones are disabled. An AI that interacts well with humans, displaying polite curiosity and delight at the sight of people will remain on while killer drones are disabled. And those that interact best with humans, due to their similar reasoning patterns, are those machines most likely to be mass produced.
- There are people who like to upload their minds into massive databases to live out their lives digitally. These people want artificial intelligence forms that think like us, because they hope to become AI themselves.
More by this Author
What is the difference between information technology and computer science?
What is Application Centric architecture? What is Data Centric architecture? What is Message Centric Architecture? And how do they compare to each other?
What are third party rights in contracts? When does a third party have obligations or rights per a contract?