We are accustomed to receiving advertisements on various websites and social media platforms based on our search history. Does it ever happen to you that you talk about a product while on the phone, and voila! You get an advertisement for it on your social media and websites that you visit! It would be exciting to unravel the secrets of this sorcery, but alas, it’s no sorcery, just AI, short for Artificial Intelligence – the new buzzword, and tech, which is no less exciting and interesting than sorcery!
Well, AI might be the buzzword now, but it is not an overnight success, it took decades of research by computer researchers to reach this point. The earliest research in this area dates back to 1950, with the famous Turing test proposed by Alan Turing in his research paper, “Computing Machinery and Intelligence” according to which a computer can be considered intelligent if it can mimic human responses under specific conditions. This was followed by numerous machine learning algorithms (which enable a machine to learn and improve from experience) that fueled the developments in AI.
To mimic human behavior, a machine needs to think like a human, and in humans, the brain which is composed of neurons does it for us. Thus, began the development in the field of Neural Networks, starting with Perceptrons and eventually Deep Learning, which consists of Artificial Neural Networks (ANN), Convolution Neural Networks (CNN) for image recognition, a Recurrent Neural Network (RNN) for text processing, translation, sentiment analysis, and improvisation of RNN, Long short-term memory (LSTM) to retain the learnings just like a human brain, although for a short period. In addition, Generative Adversarial Network (GAN) was developed initially to generate synthetic data for training datasets for machine learning models based on existing datasets. One of the most recent and the ‘hot conversation topic’ is the Large Language Model (LLM), a trained deep learning model to understand and generate texts in a human-like fashion.
Owing to the availability of high-end Graphical Processing Units (GPUs) and other hardware advancements, the development in the field of AI has been fast-tracked in this decade. We have already adapted ourselves to the use of mobile phones with the internet, computers, and other electronic devices in our everyday life and have made ourselves the users of AI products before realizing it. For example, the doodle recognition game by Google, Quick, Draw! which was released in 2016, used our hand-drawn doodles to train its neural network for doodle recognition amassing an enormous amount of data over time. The first humanoid robot, Sophia developed by Hanson Robotics, was an internet sensation in 2016 because of her human-like expressions and engagement in conversations.
The progress in the field of AI has also led to innovations in robotics and has seen robot involvement in warehouse operations, factories, and other hazardous environments unsuitable for humans. Research is also in progress to develop miniature robots for assisting during natural calamities to locate lives buried under debris, robots for cleaning manholes, and other applications that will benefit humankind. Let us not forget the most-loved and impressive robot dance on the internet by Atlas and SpotMini of Boston Dynamics. Adding to the list of wonders of AI would be self-driving cars that soon will be occupying our streets.
While these might seem far-fetched, numerous existing AI applications have the potential to optimize our daily tasks. For instance, virtual assistants like Google, Alexa, and Siri help us to manage, seek help, and keep track of our activities. Midjourney and other image-generating platforms rely on GANs for generating synthetic images, posters, and videos. Applications like Grammarly help us in our content curation game with grammar suggestions, and the renowned Chat GPT by OpenAI addresses writers’ creative block, generates quick content, provides rephrasing services, and assists developers with coding, among many other functionalities.
With our increasing reliance on AI in critical fields like healthcare for preliminary diagnosis of diseases, analyzing test reports, etc., it becomes crucial to have unwavering trust in the results generated by AI models. This can be enabled by a field within AI, Explainable AI, that’s of immense and immediate importance. Explainable AI encompasses a set of methods and processes that explains the reasoning behind the outputs produced by the machine learning algorithms, which is otherwise considered as a ‘black box’. This will contribute to making the AI models free from biases ensuring greater transparency and accountability.
As AI unlocks endless possibilities, it is now a well-known fact that our data is constantly at risk of compromise since data is the fuel for AI. Based on our internet behavior, a virtual model is created with our details such as age group, household income inferred from spending habits, education, industry, technology area, employer size, and even homeownership status that assists in targeted marketing for brands. The virtual assistants that play the role of verbal search engines in our mobile phones are trained to catch the ‘wake words’ such as, “Okay Google” or “Hey Siri” when in use, and they periodically sample a small portion of our conversation to train themselves with our voices, a permission we may have granted at the time of installation without much thought.
In this technology-driven era, do we still have the privilege to maintain privacy and safeguard our data? The answer is no. Furthermore, if humankind succeeds in creating a complete mental and emotional clone of ourselves, and if these clones become accessible to all in the future, will we despise ourselves? Or will it inspire us to become better individuals? Alternatively, will we channel our ‘ideal personality’ to our clones? The latter seems more in line with human nature.
References:
[1] History of artificial intelligence (AI)? | Tableau
[2] Computing Machinery and Intelligence by Alan Turing | Oxford Academic
[3] Types of Neural Networks and Definition of Neural Network
[4] What is Explainable AI (XAI)? | IBM
[5] Quick, Draw! by Google
[6] Interview the humanoid robot named Sophia (Full) | CNBC
[7] Do You Love Me? dance by Boston dynamics robots
[8] Is my phone listening to me? Yes, here’s why and how to stop it – Norton
