Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.
Can technology be sentient? Since the first artificial intelligence (AI) program was written in 1951, researchers and technology professionals have worked tirelessly to develop highly sophisticated AI programs. One of the early pioneers of this type of technology was Alan Turing, an English mathematician and computer scientist. Turing understood that as humans, we combine information that is available to us with reason to make decisions. He theorized that because it was possible for humans to get to a logical conclusion using these methods, it was conceivable a machine could do the same.
Around the same time, we also saw popular culture use the emergence of AI and robots to create a new class of villains — robots with human intelligence that could feel, emote and connect as humans do, taking over the world. It resulted in a fear of advanced technology that has persisted over the last seventy years in movies, pop culture, and books.
Outside of popular culture, scientists and engineers were actively working to develop smarter and more advanced AI programs. With Turing believing early on that AI could be programmed to make decisions, it opened the door for scientists to ask a very critical, albeit philosophical question, could AI ever become advanced enough to become “sentient”? Whether this would be beneficial or dangerous is mostly up to individual interpretation, but despite recent headlines, sentient technology isn’t here yet, and it won’t be during our lifetime.
That’s because AI and machine learning (ML) are still in their infancy and there are great strides to be made when it comes to optimization and innovation. We have mastered many of the building blocks necessary to create sophisticated AI systems, but we can’t build the full sentient being yet.
MetaBeat will bring together thought leaders to give guidance on how metaverse technology will transform the way all industries communicate and do business on October 4 in San Francisco, CA.
Society has many different ways of defining the term “sentient”
To really understand what it would take for the technology to become sentient, it’s important to break down the philosophy behind what Western civilization defines as “sentient.” We also need to differentiate between a trained technological device and a legitimately autonomous decision-making machine. Colloquially, we define sentience as a being that is conscious of oneself and that has agency and autonomy over their decisions.
In the early 1900s, Turing studied this idea, and what it means for something to have consciousness. As a result of his research, he developed a test to determine if a machine has human-level consciousness. The test found the AI to have human-level consciousness if a human was unable to differentiate whether they were communicating with a machine or a human.
Sounds simple, right? It’s a little more complicated than that.
For example, if you are speaking with someone who works in customer service over the phone or via an online live-chat service — a bank teller, for instance — by asking questions and interacting, we can assume they have self-awareness and consciousness. This is because they are listening to us and are able to react and respond in a way that provides meaningful solutions to our problems. Sometimes, when people interact with folks who work in customer service, there may be emotions expressed such as anger, joy, and fear that a human is able to pick up on and react to. If a machine is able to handle the same function of listening, responding, and detecting emotion in a meaningful way, how does this affect our definition of consciousness?
Sentient technology: Mirroring human interaction
AI is bound to mirror human interactions because humans are programming the software to handle functions that a person would normally do. As a result, they have implanted some of their own biases into the AI they are creating — which is a completely different story. Take, for example, chatbots. This type of technology removes the need for human employees to fill the office spaces of call centers, responding to and routing customer queries to the correct person. But the technology is built to respond and interact in a conversational format that is followable to the human on the other end of the line, or helps the person calling to get the answer or task they need to be done.
As AI continues to become more advanced, it is bound to become more complex. That being said, just because something can handle complex tasks, does not mean it is sentient. Today, AI can perform a multitude of tasks because it’s been trained to do so — talking to us, performing real-time translation, powering autonomous vehicles.
This is possible not because the AI is making decisions, but because the machine or software is following the set of rules and codified information a human has installed. It’s also worth pointing out that in most of these situations there is still a human in the loop component, and the AI is not acting independently.
The emergent phenomena
AI needs a set of rules to follow and a human to determine those rules. One example of how these rules take shape is the idea of ‘The Emergent Phenomena’ in technology. This can be defined as the appearance of something new and unpredictable in the process of organic evolution.
This means that even if a machine isn’t specifically programmed to do something, due to the training it’s received and the wider context it’s operating in, it may be able to perform certain tasks and operations relatively unprompted, which is a natural progression in the process of developing AI.
This doesn’t mean that the machine is sentient, though. Rather, it represents the achievements of current technological advancements in improving systems to help IT teams minimize time spent doing tedious tasks that a machine can be trained to perform. It’s all about the degrees of freedom or limitations that humans build into the system. This idea of AI potentially teaching itself to do things tends to be where the sensationalized Hollywood-inspired fear comes from, as we imagine machines taking over the world.
Will AI ever become sentient?
While it’s not fair to put a stake in the ground and say sentient AI will never be possible, it’s more realistic to think this type of technology is hundreds of years away. We’re just at the very beginning of what’s possible with AI, and while the idea of sentient AI is intriguing, we have to master the art of walking before we can run.
If and when it does happen, it will pose massive philosophical questions for the broader community. If a machine is conscious, do we extend human rights to that machine or access to an attorney in the case of Google LaMDA? As of right now, we have a long way to go in terms of perfecting general AI before we can even begin to think about or develop sentient AI.
As AI and ML continue to be developed and improved, we will certainly be able to enhance customer and employee experiences and minimize time developers spend perfecting the individual building blocks of technology as the bigger picture starts to come together.
Although the idea of a science fiction AI taking over the world might be a great plot for a movie or podcast series, we can rest assured that technology is a friend rather than a foe. Widespread adoption of AI into everyday lives will normalize it and build trust from a human perspective and remove the layers of fear left of years of sensationalist sentiment toward ML.
And while we might think our Siri or Alexa is mad at us, she is definitely listening, but we can rest assured that she isn’t a sentient being.
Adam Sypniewski is the CTO at Deepgram.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!
Read More From DataDecisionMakers