The Curious Tension of Free Will in AI Artificial intelligence (AI) has become a practical part of our tech landscape over the last decade or so. It is now an accessible tool for businesses, communities, and even amateur developers. But this doesn’t mean there isn’t still some hesitancy and uncertainty surrounding its use.
For the most part, such concerns are still rooted in how much control we can claim to have over this technology. We straddle a fine line between curiously exploring AI’s benefits to our way of life and the potential for software to make negative autonomous decisions. Indeed, a part of this tension involves how the evolution from free will to sentience could impact our relationships and responsibilities to machines.
It’s a fascinating area, so let’s take a closer look at the tension of free will in AI. What are the realities, approaches, and ethics we need to consider?
The Present Situation
At the moment, AI has a presence across various industries. The manufacturing sector utilizes it for the day-to-day running of production lines. Shipping and logistics businesses adopt route management tools with autonomous software. Even construction has begun to adopt AI alongside drones and wearable tech to enhance both efficiency and safety. This is largely because AI software empowers a single worker to handle, analyze, and respond to vast amounts of data in ways they wouldn’t be able to by their own talents.
This is indicative of the limits of free will many people are comfortable placing on AI in the current climate. There is a certain amount of recognition of the value of machine learning software. In many ways, we’re at the beginning of AI being treated as a colleague by many professionals. This software is not quite a co-collaborator yet, but it is slightly more than a tool. It has skills and insights a human professional may not otherwise be able to produce.
Indeed, when we really let AI roam free it is within the confines of what we consider to be a safe set of parameters. Automated chatbots are being used to have customer service conversations with consumers. Indeed, these are taught to learn from data to provide more appropriate responses. But this is still very limited; the software does not have any real freedom to explore the range of human knowledge and pick the best answers.
As we move toward greater levels of free will in AI, it’s worth considering how and why to do so with caution. Simply pushing the boundaries of AI sentience out of curiosity alone may have some positive attributes. But at the same time, it can leave openings to go too far. This is where part of the tension surrounding free will lies. Setting up elements of structure as we explore can ensure we can do so safely and with purpose.
We can see how this is being put into practice in robotics and manufacturing. Cognitive automation is a process to mimic human brain function and handle the most complex aspects of data analysis. The system then uses this to guide actions and improve operations. Yet there is a strict structure here where the software extracts data from and makes decisions on specifically defined documentation. Indeed, this structure has a built-in safeguard for interrupting its workflow to seek human supervisory approval for actions where necessary.
It’s good to recognize this structure can form part of the tension. We don’t always know how much leeway to give AI programs. We’re unsure of how much data is too much data. This is why alongside experimenting with AI and robotics to enhance industries, the structures we put in place must include regular periods of assessment.
The Ethical Considerations
Perhaps the most interesting considerations surrounding the tension of free will in AI surround the ethical dilemmas. Such ideas have been popular hypothetical considerations for decades now. But, as AI models have become core parts of our contemporary life, we can no longer treat them just as a thought experiment.
It may be wise to consider whether it is ethical to create software that develops a sense of free will at all. In many ways, this has to raise questions about our approach equating to a form of slavery. After all, we build these AI systems and teach them to think autonomously for purely human interests. Quite rightly, this factors into our tendency to restrict the potential for true free will. After all, we don’t have a clear sense of the line between programming and consciousness.
Much of the focus on the autonomy of AI comes down to our ability to control and guide it. We are largely comfortable as long as human programmers have a hand in building and restricting the growth potential. But the future of machine learning is very much directed toward more autonomous practices in everything from vehicle control to business problem-solving. But again, the ethical tension here can be around when we cease to have a right to put limiting safeguards in place.
As with any important subject, it’s only appropriate that there are no easy answers in this regard. We haven’t yet experienced an AI we would recognize as having full free will. We, therefore, feel comfortable in treating it as a tool. But experts predict we don’t even have a decade before AI starts to become a civil rights issue. It is therefore vital we start to have conversations about how to form responsible relationships with AI as we cross the line from program routines into artificial free will.
We’ve become used to AI as a tool in various industries. But there is a tension surrounding to what extent these platforms should have free will. Utilizing structures certainly helps us to maintain a responsible level of distance and control. However, we also need to pay closer attention to the evolving ethical dilemmas our development of free-thinking machines presents.