One AI for us all



I would like to paint a picture for you of a not-too distant future — a world where all the apps on your phone are merged and reduced into one singular app that oversees everything in your life. It exists in the cloud — accessible from every location: any computer, any airport, any self-driving car. It recognizes you using multi-factor authentication. It manages your life with the efficiency of a personal assistant, it integrates knowledge and provides insights with the genius of a Nobel Prize winning professor, it knows your thoughts, fears and concerns with the intimacy of a therapist, and it predicts your behavior and needs more accurately than you can, yourself. What once seemed like science fiction is here, now.

The Past

So how did we get here? One of the biggest myths in the zeitgeist is that the current wave of AI technology is new. The fundamentals of this technology have existed, prominently, in the developer community for the past two decades. They have silently powered our search bar, our social media feeds, our gaming apps, and our movie and music recommendations.

One of the most important advances in AI was the development of machine learning, a type of AI that allows systems to learn without being explicitly programmed. Machine learning systems are able to learn by analyzing data and identifying patterns.

Machine learning has been used to develop a wide variety of AI systems, including: natural language processing (NLP) systems that are able to understand and process human language (think: machine translation, speech recognition, and text summarization); computer vision (CV) systems that see and understand the world around them (think: object recognition, image classification, and video analysis); and recommendation systems that suggest products, services, and content to users (think: online shopping, music streaming, and movie recommendations). These AI systems form the bedrock upon which the house of AI is built.

The Present

The brave new world is closer than we can imagine, because in some ways it already exists. It has been ushered in through the integration of already established technologies like virtual assistants, smart devices and chat bots.

Bots like Siri and Alexa (examples of Artificial Narrow Intelligence or ANI) listen intently, overhearing virtually all your conversations and can research and look up questions, and follow through on simple commands from buying stocks to restocking the milk in your fridge. The Apple watch (and other devices like it) integrate into every aspect of your life, from tracking your sleep and your heart rate, to giving you directions based on your current GPS location, to texting back your spouse or emailing your coworkers. It can remotely control the locks on your door, the temperature in your house, pay bills and manage your calendar.

Most recently, some people are turning to AI for mental support: In April, Snapchat released free access to My AI, the app’s built-in artificial intelligence chatbot. Millions have flocked to use the AI chatbot and some teens are seeking advice, engaging in therapy-like conversations, and finding comfort or discomfort in their human-like conversations with the AI.

If we were to imagine AI’s innovation curve like a hockey stick, we are right at the inflection point before exponential growth. In March, OpenAI launched an API platform that enables other tech companies to incorporate GPT-4 into their apps, further building on it. Similar to the way that Apple and Google helped usher in a new era of app driven technologies through their ‘app markets’, OpenAI enables plug-ins, driving an entire new suite of applications that use ChatGPT to advance their applications. Reciprocally, these new applications will enhance ChatGPT’s abilities through (limited) access to the data from its engagement with users. But the biggest driver of growth won’t come from human developers — it will come from within AI itself. Like a baby emerging into early childhood, AI evolves, learns and develops the more it gets exposed to humans.

The (Not-So-Distant) Integrated Future

We are quickly expanding from generative artificial intelligence, existing in silo’ed platforms like Chat GPT-4, Bing, Ernie, Dall-E, and Bard, towards a future of fully integrated artificial intelligence. In the ever-forward march of innovation, we are advancing towards a world of artificial general intelligence (AGI) where computers can do anything the human brain can do; where machine intelligence matches (then supersedes) human intelligence. How distant is this future? Google’s DeepMind CEO predicts we will reach this precipitous point in just a few short years.

On a philosophical level, it seems inevitable. Unlike us, AI systems can scan the entire database of human knowledge that has ever existed across the entire internet in every language, in a nanosecond. Unlike us, AI systems can process gargantuan amounts of data, identifying patterns that would be impossible for humans to see. Unlike us, AI systems can learn, instantaneously, from one another. (When one machine learns, they all learn instantly, sharing their insights across the entire network.) Unlike us, AI can “parallelize” its processing, and distribute work, reducing what would take entire human lifetimes to complete, down to a matter of seconds. And unlike us, AI needs no food, no rest, no compensation, and no consolation.

AI is constantly learning and developing — moving at a pace so rapid that it is difficult for the human brain to comprehend, predict, and even keep up. It is rendering obsolete entire companies, industries, political systems and ways of life. It also has the potential to exacerbate already existent social and economic disparities.