Artificial Intelligence: The 5th Generation Superpower

It is truly amazing how advanced we humans are – how we have been given the ability to think and reason, which differentiates us from most other animals. Human beings are supposed to have one of the most advanced brains (Yeah well, some researchers argue that the Minke Whale might be smarter than humans, in some respects!) – but we hardly questioned it, because we thought, “Well, that’s just how it works, right?” Most animals have a brain in proportion to their size, however the human brain is nearly 6 times larger than expected! As many hypotheses suggest, this is largely due to the complex nature of the problems we encounter and solve - particularly social problems. Solving a social problem requires you to anticipate how your “friends and foes” are going to react to every move that you make, and this naturally is not an easy task. I mean, who else, or what else, could possibly possess a cognitive intelligence that could rival ours?

Well, like I said, humans are capable of remarkable things. But we weren’t fully aware of its extent, were we?

AI is supposed to be the ‘5th Generation’ of computers, but the research had begun at a time even before the 2nd generation – at a time when computers were still 100 feet long! It was about 6 decades ago, in early to mid-1950s, when scientists and researchers from various fields began exploring the possibility of creating an artificial brain!

Fast forward to present day – when AI, Machine Learning, Data analytics etc. have become some of the hottest trends in the market! The extent of innovation these days in AI is truly incredible. Ever imagined a courtroom filled with AIs engaged in legal battles? Well, only a few years ago, a Stanford graduate – Josh Browder – had developed the world’s first Robotic Lawyer!

Joshua Browder – founder and CEO of

DoNotPay (a legal services chatbot).

That’s not all – we also have AI-produced music and AI-created books these days! ‘1 the Road’ – a 2018 novel - was written entirely by an AI! Here, the AI had narrated its own cross-country road trip from New York to New Orleans.

‘1 the Road’ by Ross Goodwin

The point I am trying to make here is that AI is not only becoming smarter – it is also beginning to show signs of human-like creativity!

It’s only natural, that with such increasing intelligence and adaptability of machines, there are equally increasing concerns about human safety and viability. We constantly hear about potential threats such as unemployment, public security and privacy issues being posed by AI. Think about it, our life nowadays is practically on the internet!

But, we would be mistaken to think that there is no solution to this. For example, multiple studies have been conducted by researchers on developing privacy-preserving AI techniques like Federated Learning. Federated learning was first introduced by Google in 2017, and has recently been adopted by Apple for its SIRI voice assistant.

If you use an iPhone, you would know that now the voice assistant on the phone will “wake up” when you say, “Hey Siri,” but NOT when the same phrase comes from others, say your friends or family. To implement this voice recognition system, you might think that Apple would need to collect a lot of your audio data. Surprisingly, it doesn't. Instead, it uses Federated Learning, that allows it to train multiple copies of a voice recognition model across all its users’ devices, using only the audio data available locally. It then sends just the updated models back to the central server. In this way, the raw audio of users’ Siri requests never leaves their Apple devices, but the assistant constantly improves in identifying the right speaker.

Federated Learning concept (Source: OpenMined Blog)

Apple has also combined this with another technique called Differential Privacy, since iOS 13, to add an extra layer of protection. This technique injects a small amount of ‘Noise’ or inaccuracies into the raw data before using it for training the ML model. It basically changes some of the individual data points while maintaining the overall results/patterns. This makes it exceedingly difficult for malicious actors to reverse-engineer the original audio files from the trained model.

Again, there is a lot being done to handle the concerns over job losses due to implementation of AI. A growing number of companies these days, including Accenture and McDonald’s, are implementing AI in their organizations without losing a single human employee. These employees that are being replaced are being up-skilled to work in other areas of the company. Accenture replaced many of its entry-level data entry employees with AI systems but then trained them in areas like business strategy and moved them to other departments where their skills could be better used. By allowing AI to contribute in areas where it thrives, humans have the chance to upskill themselves and even find better opportunities.

As Jacob Morgan, a famous Keynote speaker, best-selling author and futurist says, “AI could be the boost that takes your career to the next level and encourages you to learn something new. There are certain skills that are uniquely human and can’t be duplicated by a machine. As AI takes over many of the mind-numbing tasks humans used to do, it frees us to focus on what makes us human. We get to build relationships and be innovative.” Honestly, it is all about having an open mindset and thinking from a different perspective.

Of course, there is still a lot left to be done in dealing with such challenges. It is an ongoing process. We can just hope for the best (and probably help in doing so, if you are a data scientist)! As Stephen Hawking rightly said:

“AI is likely to be either the best or the worst thing to happen to humanity”

I hope, and I am positive, that it will not be the latter!

---

Harsh Doddannavar is an undergraduate Computer Engineering student at Nanyang Technological University, Singapore, aiming to specialize in Data Analytics and Artificial Intelligence. He has always had a keen interest and passion for the world of AI, right from his high school days. With an aim to make the best use of the available time during quarantine, and develop something useful and informative for the community, he partnered with a high-school friend on a startup venture - ‘The Polling Booth’. It is an online polling platform that makes use of data analytics to understand and highlight real public sentiment. Their mission is to help people find out what others truly feel about different socio-economic issues, and not get overwhelmed by the sea of information and news out there.

It currently operates on a B2C business line, conducting community polls and preparing reports based on the analysis. You can also subscribe to their newsletters, to receive a key summary of their reports, straight to your inbox! To find out more about them and receive updates, visit their website and follow them on their Instagram and LinkedIn Pages!