The industrial revolution of the information age

One of the main characteristics that made us so successful as a species was the ability to adapt and optimize. This phenomenon can be observed throughout the entire lifespan of evolution if we look at the history of how we have changed from a primitive society into a very complex civilization. Both from an engineering and behavioral perspective, one can attribute the majority of today’s progress to the elementary skills of a person for finding an optimal solution for a problem at hand and learning to perfect that solution. From ancient history, we know that people in the stone age came together as social herds and realized that creating reusable tools such as spears and arrows was much more effective in their hunting success. This basic notion later became an underlying factor for survival, increasing their chance by several orders of magnitude. It was such behavior that drove and inspired modern people to migrate from limited production into industrial manufacturing.

Looking to the past, one can realize that progress has never stopped and always came up with better alternatives to existing problems. For instance, mankind had realized that horses were a much better means of transportation, but very inefficient; thus, coming up with vehicles. The same applies to a railroad, cargo, airplanes, and everything we take for granted. Such tendencies allowed us to come up with the science of discrete domains, which allowed us to create better technology for improving our lives. Ironically, in the modern world, it is significantly hard to identify the boundaries between whether technology is changing us as a species or if we as intelligent beings are changing technology. With the hype in the world wide web and the transition from industrialization into the information age, we now realize that there is a whole new dimension to our progress. This new insight made us realize that not only we can outsource the mechanical labor to machines, but also the intellectual tasks. This new form of problem-solving is called artificial intelligence (AI), which has the power of not only improving our lives but also modifying us as a species. One can look at the picture portrayed by the cinema industry, where machines put end to biological species, or simply rely on the ongoing progress in the field and summarise the benefits this new technology is bringing.

From mathematics or computer science, perspective AI can be referred to as a superset of statistical methods, machine learning, deep learning, reinforcement learning, or a combination of several methods, but for the sake of simplicity and single point of reference, all the mentioned methods are referred to as AI.

One of the positive impacts that AI is having is related to medicine and genetics. Often creating a new drug requires many complexes, as well as expensive steps, and the majority of the designed products, never make it to consumers. This could be changed in the new era of AI, where the success of a product is dictated by running several simulations in the lab. To demonstrate, it is considered that DNA carries our primary genetic information, and its transcription to RNA allows creating new proteins. In other words, there is a sequence that is converted into a structure that determines a function. Knowing how to create structures and how to predict the folding of RNA into structures is very valuable, as it allows for more controlled experiments in molecular biology, in addition to opening up possibilities to create microstructures for medical or other purposes. More precisely, the challenge lies in finding a structure that is formed by RNA and has Minimum Free Energy (MFE). MFE is considered to be the most stable structure for a given sequence and finding it computationally is an NP-hard problem [1]. Therefore, relying on reinforcement learning methods to solve this problem is reasonable. The horizon of AI in medicine extends beyond genetics and could make it possible for more affordable and ubiquitous diagnosis or continuous monitoring of one’s health.

It is no doubt that AI is transforming almost all industries and is already impacting millions of lives daily without us even realizing it. For instance, it is almost impossible to navigate large cities without using services such as Google Maps, which uses machine learning to recognize building edges, shapes, street numbers and create a map of a city [2]. Another example is Google Translate, which uses the Neural Machine Translation model to translate between languages [7]. Other examples include face ID on smartphones, search engines, digital assistants, smart home devices, banking services, recommendation systems, and the list goes on [4]. Perhaps the most exciting application of AI is demonstrated by Irina Rish, an associate professor at Université de Montréal and a research staff member at IBM, who explores the bidirectional relation between “AI for neuroscience and neuroscience for AI”. In her presentation given at “The Machine Learning Conference” and as a guest lecturer in Aalto University in 2018, she summarises her work and claims that AI can be used in neuroscience “to recognize mental states and identify statistical biomarkers of various mental disorders from heterogeneous data” collected by “neuroimaging, wearables and speech” [5, 6]. On the contrary, neuroscience can be an aid, reference, or inspiration that helps in building a better, more efficient, and optimal AI. Such claims are also stated by Demis Hassabis, a researcher, CEO, and co-founder of DeepMind [3].

While some researchers are trying to understand, clarify or explore both human intelligence and AI, others are trying to integrate and use intelligent machines within a biological brain. Such work is pioneered by Elon Musk and the team at Neurolink. The underlying philosophy behind such an ambitious project is best explained in the article by Tim Urban: “Neuralink and the Brain’s Magical Future” [8]. To clarify, the article argues that human’s evolutionary progress resembles logarithmic growth, our communication speed is limited and if we embed a brain-machine interface (BMI), then we could fictitiously withstand the era of GeneralAI. This new form of augmentation of the human brain could completely change us as human beings, but its initial positive impact could help millions of neurological patients.

In brief, there are many aspects of AI that are unclear at the moment. From an optimistic perspective, AI could help improve our lives. However, we should always keep in mind the fairness and biases, as well as transparency and trustworthiness when building artificial intelligence, as it is just another reflection of human beings.

References

Data Scientist and Machine Learning Engineer at Plan Brothers. maksad.github.io