Potential of AI Leading to Adverse Disruption

Since the first machine was built, we have always imagined a possibility of a future where we co-exist with intelligent machines that roam the streets from their own free will. This vision is the premise in a number of books (many of the sci-fi books by Isaac Asimov), films (The Matrix, I, Robot) and TV shows (Futurama, Westworld).

These visions were fun to imagine and we never really considered the consequences of such a reality since it was all hypothetical. However, this is becoming less fiction and more real. Technology is advancing rapidly with many institutions targeting the Artificial Intelligence (AI) sector.

In fact, there are many VC funds being raised to target this market — for example, Andrew Ng, who formerly founded and led the “Google Brain” project and served as VP and Chief Scientist of Baidu, is raising a $150m AI fund. In the last few months, Google has confirmed their “AI-first” fund, Gradient Ventures, Microsoft Ventures started its own AI fund and many others have announced AI related funds. With all of this capital helping to accelerate AI startups raises many questions that we never thought necessary to answer such as “Will robots steal our jobs?”.

Before we get into the potential implications of the AI revolution, I just want to make a clear distinction between AI and machine learning, which are used interchangeably by many but are not actually the same thing:

• AI: machines capable of intelligent behaviour, where they carry out tasks in a “smart” way
• Machine Learning: computers that use algorithms to process a large amount of data, learn from it, and then take an action based on this analysis — essentially it is getting computers to act in a certain manner without telling them explicitly what to do

So AI is essentially building a smart machine whereas machine learning is a process that enables AI and makes the machines truly intelligent. Now AI can be generally classified into three categories:

• Artificial narrow intelligence (ANI): AI that is skilled at one specific task
• Artificial general intelligence (AGI): AI that is considered as smart as a human and can perform a range of tasks
• Artificial super intelligence (ASI): AI that achieves a level of intelligence smarter than all of humanity combined. Nick Bostrom, a leading AI thinker, defines superintelligence as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills.”

Currently, we have achieved ANI and it is all around us from predictive search suggestions on Google search to “recommended for you” products on online retailers to email spam filters. The AI revolution is proceeding from ANI, through AGI, to ASI, especially with the progress we are making in machine learning — what could happen at the end of this road is the question everyone is worried to answer.

A great book that essentially sums up the happenings in AI thus far and has a look into the future is “Surviving AI: The promise and peril of artificial intelligence” by Calum Chase, who is a regular speaker on artificial intelligence and related technologies and runs a blog on the subject at www.pandoras-brain.com. The book considers the challenges and potential benefits of AI as well as looking at the history, current technologies and future developments of AI.

It made me more aware to the AI around us, such as retail shops having products in stock by using AI to ingest huge data feeds and run algorithms to analyse and predict when, where and what we will collectively buy. In fact, I never really consciously thought about it but I use Amazon and Netflix to watch films and TV shows and they use AI to suggest programmes that I would like based on what I have watched in the past. It is crazy to think about how integrated AI is in our everyday lives.

It also gave me insight into the issues with developing AGI and ASI, which I believe are areas of concern that we need to resolve before we reach this stage. In my opinion, there are two significant challenges — one that is far in the future and one that is happening currently.

1. Killer robots

Let’s talk about the one set in the future first. Killer robots! If you watched The Terminator, you will know exactly what I’m talking about. Until very recently, not many people really thought about AGI and ASI. This changed with the publication of Nick Bostrom’s book, “Superintelligence” (also on my reading list) and the acclaim it received from Elon Musk, Bill Gates and many others. In fact in January 2015, Stephen Hawking, Elon Musk, the co-founders of DeepMind, and many artificial intelligence experts signed an open letter on artificial intelligence calling for research on the societal impacts of AI.

The worry is that once we develop AGI, the machines at the start will do a lot of jobs for us but they could soon become too smart… communicating with each other, they could develop into an ASI. This is a serious concern — look at Facebook’s recent shutdown of its AI project.

The experiment set by Facebook was to have chatbots negotiate with each other over a trade e.g. swapping balls. They were instructed to negotiate and improve their bartering as they proceeded. The AI progressed to talk to each other in a language they each understood but looked like gibberish to humans. Some of the negotiations in this language actually succeeded! The project was abandoned because the AI were not doing the work required, as it wasn’t useful for humans since the language was incomprehensible. Regardless, this outcome was intriguing since it showed that allowing AI to have an unconstrained environment resulted in a situation that was not beneficial to humans.

Taking this one step further, what if ASI develops in an unconstrained environment and decides that humans are inferior? These machines could be able to process information infinitely faster than humans and visual every reality, every possibility within seconds — they might decide that humans are adversely impacting the planet and exterminate us or worse, imprison us (or maybe even put us all in a coma) so that we cannot cause further harm to each other and our surroundings. Therefore, we must make sure boundaries for AI exist before we let them loose.

2. Robots taking our jobs

Now the killer robots scenario is in the distant future and has significant consequences for mankind but in most peoples’ minds currently is a more pressing challenge that is happening right now. The big concern is that robots will take our jobs — according to PwC:

Up to 30% of UK jobs could potentially be at high risk of automation by the early 2030s, lower than the US (38%) or Germany (35%), but higher than Japan (21%)

The risks appear highest in sectors such as transportation and storage (56%), manufacturing (46%) and wholesale and retail (44%), but lower in sectors like health and social work (17%)

On my street earlier this week, I saw a Domino’s and Starship Technologies pizza delivering robot, accompanied by a human helper but with the intention to be standalone in the future. The future powered by AI is getting there and the speed of this revolution is increasing by the minute! Calum Chase in his book said that we might get to a point when a majority of jobs will be performed more effectively, efficiently or economically by AI than humans — this is the concept of economic singularity and we do not know what is on the other side of this point. This scenario is beneficial for society but has negative implications for people whose jobs are taken by the machines.

What can we do to combat this? Well this is up for debate and in my opinion a solution is to simply educate ourselves and those around us so that we can gradually adapt to the technological changes around us. I admit this is easier said than done and there are other such propositions such as the robot tax suggested by Bill Gates. However, all proposed solutions have problems — for example the robot tax could be seen as a tax on innovation, and there’s a bigger problem of defining exactly what is a robot and when it takes your job. Looking at the pizza delivery robot, it is taking away the delivery job from humans but is also introducing a new job to take care of and maintain the robot. Therefore, it is difficult to say when robots are completely removing jobs from society.

There are no easy solutions to this problem, which is accelerating and will significantly disadvantage unskilled workers. As many others have mentioned, this divide between the rich and poor will only get larger as we progress with technological advancements so we must prepare now not just for the short-term societal impacts but potentially the future of the human race!

By Abhilash Dubbaka

Abhilash Dubbaka is currently an Investment Banking Analyst and an investor with a passion for the Technology sector. He has a keen interest in FinTech and AR and VR sectors. If you have any comments, please contact Abhilash through LinkedIn or reply to this post.