AI for Good and Bad
Taking a wild guess on the future of artificial intelligence and humankind
For distant future readers
It's spring 2023…
Some will say that our civilization is doing well and extraordinary times of prosperity are ahead of us. Others predict an inevitable end in the near dystopian future. There are also supporters of the view that everything's about the same: wars and plagues alternate with better and calmer times.
I find it challenging to agree with any view. On the one hand, as an Earth and environmental scientist, I know that we messed up big time. It is difficult or even impossible to reverse human-made climate changes, fix species extinction, and decrease the pressure we put on nature. On the other hand, as a machine learning engineer, I'm watching with bated breath what's happening in artificial intelligence. I believe that some future AI-enhanced technology will allow us to repair the damage done.
There are multiple possibilities to utilize AI to shape a better future or, on the contrary, do something terrible. I'm sure there are people actively working on both solutions now. After all, there are many recipes for applying AI for good and evil. Just pick up some sci-fi books and let your imagination run wild.
I have always been tempted to take on the challenge of predicting what will happen in the future. Hence the idea for this article. In it, I will attempt to figure out how artificial intelligence will affect people and nature in the future. I write it primarily to see how my worldview will change in a few years.
Disclaimer: it is more a collection of loose thoughts than serious reading. However, I invite you to share your thoughts and ideas. Especially to prove me wrong.
You, the reader from the not-so-far future, will either congratulate me on my accurate predictions or rejoice at how wrong I was!
AI for Bad
We'll start with the difficult topic of AI harming future humankind. For us AI experts, it will be essential to counteract this phenomenon. The sooner we start, the better.
Fake science
It is not faith or good intentions that will help us solve problems on a global scale but high-quality scientific research. This quality will soon depreciate.
Don't get me wrong. I personally jumped on the generative AI and large language models (LLM) hype train, but the impact of this AI discipline on our daily lives scares me. Regarding science, I'm not worried about the mass generation of scientific articles or authors becoming similar to each other due to the use of the same writing-assist software.
I have serious concerns about data. Data set augmentation has long been used in machine learning. I am a fan of this technique myself. But what happens when we can produce complete datasets that are realistic but do not represent a real natural phenomenon?
Data collection is a highly complex process in all disciplines of science. This task generates enormous costs and is often dragged out over time. One can easily avoid the trouble of getting the data by asking the LLM relevant questions. The answers will not be real, but they will look realistic.
Are we going to take shortcuts to keep up with the publishing cycle? Soon, we will see a minor scientific revolution, where the critical value, apart from innovation, will be the rigorous repetition of experiments, not only classical peer review.
Praise the machine god
We place a lot of trust in AI. The fact is that machine learning or deep learning solutions help us a lot in automating various types of tedious processes. Often, ML/DL solutions also participate in decision-making.
AI begins to become commonplace, and our familiarity with it increases. We start to believe in it. It's almost like we start trusting a person because we've known them longer and longer. However, this is only a tool. Like everything, it can fail us in the least expected situation.
At the same time, AI is becoming increasingly difficult to understand. At some point, we will be unable to keep up with it. The only thing left will be going with the flow of technology. I wonder how quickly we'll accept that we live in a world controlled by algorithms we don't understand. How to promptly accept that we are hurt by rules over which we no longer have control? It may have already happened.
An unhappy accident
I am concerned that some AI-enhanced technology will emerge, the results of which we cannot control or even perceive. I'm not talking about a robot uprising. Let's be serious. I am more troubled with mechanisms that could significantly and negatively impact the direction in which society develops.
I know it's hard to imagine. Let it be a helpful solution initially, which introduces some social disproportions due to its long-term operation. A citizen rating system, an interesting algorithm operating on the financial market, or an application serving us entertaining content on the web. Left unattended and unmonitored for years, they can have an impact that is difficult for us to see. The rating system may introduce a new kind of discrimination, the financial algorithm deprecates specific individuals, and the content recommendation application will make us less sensitive or stupid.
Power consumption
AI has the potential to provide an incredible level of control over even the most complex processes. It already outperforms humans in many demanding tasks. Every year, there will be more and more solutions based on AI.
However, this has a cost. Today, training ML models and inference require significant computational resources. It's about more than just handling user traffic. The problem is the need to constantly retrain models. This is important for the solution to work correctly and resist data drift. At the same time, models are getting more and more complex. Working with them costs you more than just time. First of all, you need electricity.
We have a big problem with this today, and it can only get worse in the future. It will turn out that we spend energy on further amenities that serve only us, humans. We will be stuck with great working air conditioning, cryptocurrencies, and many AI models in a world with a destroyed natural environment.
Obsolescence
It is a known fact that occasionally, some professions disappear, and new ones appear. My kids are probably learning to do jobs I haven't considered yet. But I'm not sure whether it's not happening too fast. With AI, the degree of automation will increase even more. At worst, we'll leave everything to well-crafted algorithms and take the human out of the equation.
I don't worry much about people using new media, working in STEM fields, etc. There will be no shortage of work—quite the opposite. I would pay attention to people who, for various reasons, did not have a chance to receive sufficient education to participate in the change we are currently witnessing. Somewhere along the way, there are a lot of difficult decisions to make not to increase the existing social disproportion.
Loneliness
This worries me the most. Supported on all sides by AI, we will not require the help of others in performing our daily tasks. We will finally be left alone with our computers. They will shop for us, help us do our job, solve our homework, play games, and recommend movies and music. We can also talk to them about various topics and spend time together. Over time, it will improve and replace our human being companions. The ubiquitous automation will ensure that all our needs are met, and we will be alone. Creepy and disturbing.
AI for Good
So far, it has been sad and terrible. Now, it's time for some good news.
Research boost
Fortunately, we don't have to wait for it to happen. AI is already involved in finding novelty in various science and engineering disciplines. I'm especially thrilled about all discoveries that await us in life sciences. New types of cures that will make us live happier and healthier lives or substances that will prevent or even reverse pollution. Due to a breakthrough in AI, we can generate more new ideas and test them faster. In this case, AI will be a tool for future researchers to conduct more complex experiments.
Quick response
A considerable advantage of AI is efficient decision-making and the ability to work continuously. We, as humans, cannot achieve this. At the same time, AI-based systems can process much more complex data than human experts. This gives fantastic opportunities to observe various phenomena and react to them. Weather monitoring, earthquake prediction, and diseases tracking are only a few examples.
In the future, we will improve our planet monitoring mechanisms. I am waiting for a complete Earth digital twin to be created, allowing us to track and predict phenomena on our planet.
Accessibility
We are facing times of faster access to knowledge. Machine learning models are getting better and better at finding information and serving it in a form that is appropriate for us. I am eager to see how AI will help us in learning and affect the education system. Furthermore, I'm sure that soon we will have a flood of various types of AI assistants. Among them, there will probably be a few valuable ideas for supporting people with disabilities or helping in providing medical treatment.
Sustainability and growth
I have the greatest expectations and hopes for using AI in sustainable development. Utilizing AI in different types of systems supporting modern societies could help us organize individual areas of human activity to run optimally. AI will be able to cope with the high complexity of many sub-processes related to using natural resources and the fair distribution of resources. I think of AI systems responsible for monitoring and controlling the state of water and its industrial use, using AI techniques in precision agriculture to minimize the number of fertilizers used, improving the transport of goods or services, and supporting the environment and wildlife conservation activities.
Moreover, I hope there will be integration between different systems so that they do not work alone but achieve a certain level of synergy.
Conclusion
It's hard to predict what will happen in the AI world in the future. One can hope that the power behind modern technology will allow us to do a lot of good. Indeed, there is no need to wait until tomorrow, but act now, i.e., recognize and eliminate potential threats and support positive and valuable initiatives.
All ideas on AI's impact on our future are welcome. In addition, links to all exciting projects to support the positive development of AI are welcome.
Originally published at https://www.reasonfieldlab.com.
