You need to about artificial intelligence (AI) because it will soon disrupt our lives. Unfortunately, all most people know about AI is hype, hysteria, and propaganda, often based on science fiction.
Here are five things everybody must understand about Artificial Intelligence:
1. Most of what the media labels “AI” is not artificial intelligence.
To clarify, a true artificial intelligence will think and create without human help.
Instead, almost everything labeled AI these days is merely better data processing. For example, in robotic process automation digital algorithms perform a function such trading stocks repeatedly. They call it robotic because a digital robot does the work. However, a digital robot merely mimics a human action without thinking.
In machine learning, a computer or algorithm detects a pattern, and reaches conclusions based on the pattern. For instance, Netflix (NASDAQ: NFLX) uses machine learning to recommend videos based on your past choices.
Robotic process automation and machine learning rely on raw computing power; not creativity, for decision making. Machine learning algorithms decide faster than humans because they process data at more at far speeds than people. Thus, Netflix’s algorithm can go through every video in the website’s library and match titles to your past your selections.
2. Algorithms and robots do not need AI to outperform humans.
Notably, the Alpha Go Zero algorithm used machine learning to teach itself to play the complex strategy game Go.
Alpha Go Zero learned Go by studying a database of over 100,000 Go games with no human help, The Verge reports. Moreover, an earlier version of the algorithm beat the world Go champion in three out of four games in 2017.
Computer scientists thought it would take decades for an algorithm to master GO, a famously difficult Chinese game, The Verge reveals. However, Alpha Go;from Alphabet’s (NASDAQ: GOOG), DeepMind subsidiary learned to dominate the game in hours. Additionally, Alpha Go came up with new Go strategies the champions were unaware of.
Thus, algorithms can learn and succeed at complex tasks quickly. All algorithms need to learn a task is large amounts of raw data about the task.
3. Algorithms do not need AI to take your job.
For example, Walmart (NYSE: WMT) is using algorithms to monitor customer behavior, merchandise levels on shelves, and inventory at a store on Long Island. In detail, algorithms at the Intelligent Retail Lab watch the shelves and aisles through video cameras.
If shelves are empty; or near empty, the algorithm directs a human associate to restock them. Plus, algorithms can detect problems like spills or out-of-place merchandise.
Walmart’s goal at the Neighborhood Market Store in Levittown, New York, is to reduce the need for managers. The hope is to reduce labor costs and improve customer service by eliminating higher-paid managers. No coincidentally, CNBC claims Walmart is seeking ways to trim its management staff.
Walmart’s algorithms probably use machine learning to keep track of merchandise. However, retail management is not the only job AI could threaten.
Self-driving cars and autonomous semi tractors machine learning to teach themselves how to drive. To clarify, the vehicle’s algorithms examine traffic patterns, road conditions, routes, traffic laws, and other data to mimic human drivers.
The United States Postal Service, Amazon (NASDAQ: AMZN), and Volvo are among the organizations testing driverless semi-trucks. Currently, self-driving big rigs make long-distance runs between warehouses on Interstate highways.
4. You are right to be afraid of smart machines and AI.
In the wrong hands technologies like machine learning, robotic process automation, and artificial intelligence could do a lot of damage. For instance, trading robots could crash the stock; or commodities, market by mindlessly selling vast amounts of shares or derivatives fast.
A more frightening scenario is the bad guys using machine learning to enhance their capabilities. Algorithms could help terrorists make bigger and more destructive bombs, for instance.
Fraudsters could use something like Alpha Go to devise more effective scams. Meanwhile thieves could use a program like Alpha Go to plot more effective heists. Hackers can use machine learning to crack codes and security.
Finally, machine learning could make war machines like tanks, drones, missiles, torpedoes, submarines, helicopter gunships, and fighter planes deadlier and more destructive. An algorithm could analyze military history and data about past data to devise better tactics for drones or robot tanks.
The algorithm could direct a missile; or artillery fire, to the weakest point in a fortification; or a warship, for instance. Thus, machine learning could make the next war deadlier.
5. True Artificial Intelligence is probably coming, but it nobody knows when it will arrive.
Experts think an AI as smart as a human; or an “artificial general intelligence (AGI),” is possible but have not produced one yet. In fact, Futurisms’s Dan Robitzski writes; “we have no idea how to build AGI.”
Instead, all the constructs the media labels “AI” are really faster and more impressive data processors. However, we are getting great at data processing.
Creating an AGI; a euphemism for a “brain,” is far more complex than data processing. For example, an AGI will have to analyze data and reach conclusions about billions of subjects.
Importantly, the experts think AGI is possible. In fact, 98% of computer experts believe human level artificial intelligence is possible, GoodAI and SingularityNet estimate. On the other hand, the experts disagree on when we will achieve human level artificial intelligence.
AI could be here in 10 Years
Dramatically, 37% of experts think we will create AGI within 10 years, Robitzski reports. Plus, 28% of experts believe it will take us 20 years to develop AI. Therefore, 65% of the experts think we will develop human level artificial intelligence by 2040.
Under those conditions, we could build machines far smarter than human beings by 2050. Therefore, we had better discuss the philosophical, ethical, political, and practical implications of AGI now.
My prediction is that human-level AI is coming within our lifetimes. However, technologies like machine learning and robot process automation will completely disrupt our lives and economy long before AGI arrives.