Top 10 things to know about AI for 2021 – Year In Review #18 [Countdown #3]

Top 10 Things to Know About AI for 2021

We are now nearing the end of the Year In Review series. In the previous two posts, we have started the final countdown to 2021. We are comparing the 20 thoughts for 2020 that I wrote in January (opening note) to the events of 2020. In this 18th part of the series, let us review Artificial Intelligence (AI) related aspects from my ‘opening note’. And, then we can look at the Top 10 Things to know about AI for 2021 and beyond.

Initial Thoughts About AI at the start of 2020

In my ‘opening note’, I had referred to AI in four out of the 20 thoughts for 2020. Firstly, I resolved to share more AI related thoughts and to democratize AI (#2 in my opening note). I also spoke about Natural Language Processing (NLP) and had predicted more explosive developments in this space (#13). More importantly, I wrote about businesses needing to connect the dots to extract value from AI (#14). And, finally, I introduced the CRANK framework to create a mental model to extract value from AI (#15).

List of Top 10 Things to Know About AI for 2021

A lot has been happening in the field of AI in recent years and 2020 was no different. It will be impossible to do justice, if I stick only to the four pointers mentioned in the ‘opening note’. So, I will expand beyond the above 4 points, based on the events during the Year in Review, 2020. Let us look at a list of Top 10 things to know about AI for 2021 and beyond.

#1 – General AI vs Narrow AI

I tend to make the assumption that people speak about narrow AI, when they are discussing this subject. This is because the typical use cases of AI are currently restricted to narrow AI. The narrow AI models are built to solve only a specific problem. They cannot solve other problems, and are not ‘general’ enough.

However, the common man, who is not aware of the technical aspects of AI, has a very different world view. They are usually thinking of a ‘sci-fi’ view of AI. Inspiration is drawn from what they have been seeing in movies and fictional stories for over 50 years now.

So, where are we in reality and what do the top professionals say about AI?

I attended an event earlier this year, where the AI Director at Cisco made some interesting remarks. He mentioned that the current stage of development is still at the level of narrow AI. We are nowhere close to the Artificial General Intelligence (AGI). In fact, he refused to speak about AGI, since the practitioners are nowhere close to AGI. My discussions with leading data scientists and AI engineers across industries reinforces this thought process.

It is interesting to note the contrast between the perception of the common man about AI and the ground truth, as validated by the experts.

#2 – When it works, it is no longer called AI

The vocabulary is interesting on the other extreme side as well. We saw one extreme where we like to talk about the sci-fi view of AI. The other extreme is where we refuse to acknowledge the use of AI in our daily lives as AI. When things work and become a part of our daily lives, we refuse to call it AI.

A classic example is Machine Learning (ML) models in email filtering that are used to detect spam. These algorithms are a perfect use case of binary classification algorithms from the field of ML. However, try asking someone if they ever mentioned about using an AI tool within their mailbox.

My simple explanation to this phenomenon is that we love sensationalizing and dramatizing debates. And, AI is no exception.

I would much rather pay attention to people who have the domain knowledge, than go by the popular narrative.

#3 – Deriving value from AI

Continuing on the same lines, a key question relates to deriving value from the investments made in AI. Once again, simplicity beats complexity hands down.

In my earlier posts of this Year In Review series, I mentioned about Citizen Developers and acceleration of digital transformation. Some of our digital transformation initiatives required data migration from legacy systems to the new cloud based platforms. I led the projects only with a team of citizen developers. We did not have an army of backend database and other IT professionals to call upon. Thankfully, AI tools helped us clean the data, point out inconsistencies and even rectify these using the front-end tools. More importantly, these AI based tools were accessible to the business users or the citizen developers. They did not even require a complex technology infrastructure to be put in place!

Marketing pitches for technologies revolve around big ticket examples that may be out of reach for a lot of people. However, such examples demonstrate that we can derive value from AI through much simpler use cases. We often tend to overlook the latter.

Let us search for the obvious, rather than looking for the oblivious!

#4 – A framework to derive business value from AI

You might be wondering if it were so simple and obvious, why aren’t we doing this? That is a question worth asking. The answer may vary, depending on your context. Hence, I have created a general purpose framework to understand the levers of value that we can derive from our AI investments.

I had created the CRANK framework in December 2019 for a keynote address at a Data Science workshop. This year, I published the details of the CRANK framework to make it more accessible. This can be understood by (non-technical) business users and can also add tremendous value to the AI professionals.

I believe this is an important step towards democratization of AI. Or, as I said in my ‘opening note’, to democratAIze!

#5 – Citizen Data Scientist

We spoke about Citizen Developers, who are business users that create software applications for their colleagues. A similar movement in the field of AI has been gaining momentum. In this case, the business users are also known as Citizen Data Scientists.

There has been a mad rush of people wanting to upgrade their skills and become Data Scientists. I routinely receive a lot of queries from business users and technocrats who want to learn AI and Data Science. This trend is going to continue and I do feel that there is lack of a structured mentoring mechanism.

I will try and cull out pointers from my experiences to help such aspirants. It is still quite rare for a non technical management executive to learn programming and AI. And, I went from there to create AI solutions that were awarded for innovation across Asia Pacific by Financial Times.

Until I put some structure around my learning, I will continue mentoring individuals who approach me for bespoke advice.

#6 – Is AI Child’s Play?

Let us continue with the theme of aspiring AI learners. I wrote a couple of posts in this Year in Review series about children wanting to code and kids wanting to learn AI.

I hope to cover the basic building blocks of AI in more detail in the near future. This is extremely important, so that aspirants know what they are heading into. However, let me tell you that any field of study, including AI, can be approached from multiple vantage points.

For instance, children may be very good at conceptualizing ‘What If’ scenarios. This is because they have their innate curiosity and creativity intact. However, they may not have adequate knowledge of the technical building blocks that are required for complex AI modeling.

On the other hand, adults may not be as curious and creative as children. They may have hard-wired thought processes based on their life experiences and conditioning.

In some cases, AI applications can benefit immensely from child-like curiosity, but AI is certainly not child’s play!

#7 – Augmented Reality of AI

We have looked at AI from the standpoint of an aspiring data scientist. Now, let us change our focus to business. The digital tsunami has brought about a sea change in the way organizations think about and adopt digital initiatives. Businesses need to examine where AI fits into their digital scheme of things.

I created the 5 ACEs framework to help businesses navigate the digital adoption journey. The 5 ACEs framework was introduced this year, starting with 5 A’s. In 2021, we will have more Clarity for Execution with 5C’s and 5E’s as well to complete the 5 ACEs.

I have broadly covered AI and similar technologies under the “Augmentation” category within the 5 A’s. The example of AI for enhanced Customer Experience covered in this Year In Review series helps illustrate the point. AI is most effective when it is used in conjunction with humans, to augment their capabilities and intelligence.

The 5 A’s model also talks about Automation, which points us to another raging debate – Will AI snatch my job?

The short answer is – It depends.

And, the key point is that it does not depend on AI, but it depends on YOU!

#8 – AI trends and advances in 2020

Let us turn to the events of 2020, as we continue the list of Top 10 things to know about AI for 2021. In a couple of earlier posts in this series, I had shown two charts. Let me stack these 2 charts here, so that you can see the story more clearly.

The Search During the Pandemic

As you can see from these charts, the mainstream focus shifted from AI to ‘back to basics’ in March 2020.

There was a spike in searches for ‘Digital’. Businesses as well as individuals were looking at putting their Digital house in order.

On the other hand, the searches for AI nosedived in March 2020 in the wake of the pandemic. This is understandable, since the mainstream view of AI is analogous to a science fiction fantasy.

However, this is not the complete picture, since it only represents the mainstream view. A lot has been happening within the AI community, that may not be reflected in the above charts. This year saw one of the biggest advances in the technical field of AI in recent times. It was the launch of GPT-3, that can help generate human like language. This is the largest ever machine learning model that has been trained with 175 billion parameters.

As I mentioned in the ‘opening note’, disruption is happening at the speed of light. This is particularly true in the area of NLP (Natural Language Processing).

#9 – Computational Models and Risks

Let us dwell a little more on going back to basics in the context of the pandemic in 2020.

One of the fundamental premises of these algorithms is that past patterns repeat themselves. Having been through a year such as 2020, we know that we haven’t seen anything like this in our lives. This is certainly true for our lives, but is it true for the human race?

If this line of thinking intrigues you, you can listen to this 2 minute video below. Or, you can watch the full keynote address at a Governance and Risk Management event earlier this year.

There are no short cuts or fixed rules that can help to guide us in our usage of AI. While applying these models, we have to take a look at the bigger picture and understand the context. One size and one AI model does not fit all situations. In fact, AI may not be the appropriate solution in all cases.

The good news is that the human brain is smart enough to guide us to the right answers. But, first, we have to ask the right questions.

#10 – Artificial Intelligence and Natural Stupidity

Unfortunately, the ability to ask questions has been systematically decimated by the educational systems of many countries. This brings me to the final point in the list of Top 10 things to know about AI for 2021 and the New Normal.

Artificial Intelligence is merely a tool in our hands. And no tool can be better or worse than it’s user. AI certainly has the potential to solve the world’s most complex problems. But, it may fail to match a child’s curiosity. It will certainly fail to be as compassionate as a care giver. And, will probably not attempt to understand, let alone replicate, the human soul.

Choose your tools wisely. Artificial Intelligence aims to replicate the working of the human brains. However, humanity is as much about our minds, souls and hearts as it is about our brains. You may find someone who does not understand this, even after the events of 2020. You can point them to this quote from Albert Einstein

Artificial Intelligence is no match for Natural Stupidity

Albert Einstein

Year In Review

Check out the master article for the Year in Review 2020 that contains links to all posts in this series. Also, bookmark the master article on the browser to read all the 20 thoughts from 2020 and lessons for 2021.

About The Author

Rajiv Maheshwari is a business and start-up advisor, and the co-founder of From The Experts Mouth. He is a management professional with over 25 years of experience, and worked as CEO for a decade, and in leadership roles with NYSE listed companies such as Accenture and WNS.

He is a Chartered Accountant and MBA (Director’s Merit List from IIM Bangalore) and an autodidact, who is on the path of self-directed life long learning and sharing. He is a thought leader, author and keynote speaker and has developed several frameworks to bridge the gap between academia and industry.