Artificial intelligence (AI) has been heralded as one of the most transformative technologies in recent history. But like any new technology, there have been some bumps in the road. Unfortunately, many of these early setbacks have been attributed to poor understanding of artificial intelligence and unrealistic expectations for what it can accomplish. As such, it's important to recognize that the path to successful AI adoption is going to be a long one. Here's a list of some of the biggest failures we've seen thus far:
1. Chatbots (1999 - Present)
For much of the late 1990s and early 2000s, chatbots represented a major breakthrough in the field of AI. They were programmed to carry out simple tasks over the internet using natural language processing technology -- essentially meaning that they could carry on rudimentary conversations with human beings. But despite their promise, chatbots failed to take off for a number of reasons. For starters, they were extremely unreliable and often couldn't understand their human counterparts. They were also hard to train and costly to develop and maintain. It wasn't until the development of deep learning technology that chatbots started to become a viable commercial product. Today, they're used by big companies like Amazon and Walmart to automate some of their customer service operations. They're also employed by e-commerce companies to facilitate online shopping and make recommendations to customers based on their browsing history. But while these bots still have a long way to go before they can rival the capabilities of a human assistant, they represent one of the biggest success stories to date when it comes to AI adoption.
2. Google's Image Search (2001-2009)
Google's image search was one of the most revolutionary applications of AI at the time. It was capable of identifying images by looking for patterns of colors, shapes, and textures. But what many people didn't realize was just how problematic the algorithm actually was. It often failed to identify certain objects and color combinations -- sometimes resulting in hilarious mistakes like the "potato" that appeared in a stock photo of a traditional Thanksgiving dinner. It also didn't account for context or account for common photography tricks like depth of field and color manipulation. This led to numerous embarrassing gaffes where Google misidentified things like the World Trade Center and even Abraham Lincoln's face! Luckily for Google, these problems were eventually fixed and the company went on to become the dominant player in online search. But some of the lessons it learned in the process have become invaluable to the development of future AI technologies.
3. IBM's Watson (2011-present)
IBM's Watson is perhaps the closest thing we have today to a supercomputer that can think like a human. It's capable of doing everything from answering complex medical questions to assisting with the translation of foreign languages. It can even play the stock market and help you cook better meals! But it's most famous achievement to date is winning the TV game show Jeopardy against two of the game's all-time greatest champions. And just like Google's image search before it, Watson faced its own share of criticism as it gradually began to take over more and more tasks from human workers. Once again, there were reports of hilarious blunders, inaccurate diagnoses, and confusing answers. But unlike Google, which was able to bounce back and eventually overcome these challenges, Watson has yet to prove itself in any commercial setting. In fact, it's been said that it's practically worthless outside the world of academia. This is due largely to the fact that Watson is currently unable to interact with humans in a more natural manner. This means that it often has trouble making sense of its environment and can even have difficulty understanding simple commands from a human operator. But the challenge for researchers going forward will be to find a way to teach computers to think more like humans so that they can help us to solve some of the world's most difficult problems.
4. Amazon Labels Congress as Criminals (2018)
Amazon is responsible for another face recognition blunder. Its AI system was meant to detect offenders based on their facial image, but when it was put to the test using a batch of photos of members of Congress, it proved to be not only incorrect but also racially prejudiced. According to the ACLU (American Civil Liberties Union), almost 40% of Rekognition’s (the system’s) erroneous matches in our test were of persons of color, even though people of race make up just 20% of Congress. It’s unclear if it was a fault with non-white face recognition or if the training data was skewed. Both, most likely. However, relying only on AI to determine whether or not a person is a criminal would be crazy.
5. Tesla (2022)
Teslas running autopilot involved in 273 crashes reported since last year. The cause of the accident was mainly the lack of driver control while Autopilot is active. But many drivers had reported that the car failed to recognize the stoplight and ended up colliding with the vehicles in front of them. While this can be explained by the confusion that the system may have when the light changes, it can also be a human error issue as well. The question is, why didn’t the Teslas react to the traffic light? Is this an issue with software updates or just poor workmanship on the part of Tesla’s engineers? Although self-driving cars are becoming more and more common on the road today, there are still a number of obstacles that need to be addressed before these vehicles can become truly autonomous. One of the main challenges involves the need to develop a sophisticated artificial intelligence that will allow the car to recognize and respond to a wide variety of different scenarios.
If there is one thing that unifies all of these issues, it is that failure is a necessity of success, and that trial and error is the basis of all scientific and technological development. The fact that we are even discussing these challenges is a testament to the progress we have made as a society, and our continued willingness to look to the future and explore new innovations and ideas.