100+ AI Predictions for 2021 from Industry Leaders
The technological landscape today is evolving faster than ever before. 2020 was no doubt a tough year that pushed businesses across verticals to innovate and adapt quickly to overcome the challenges brought on by the COVID-19 pandemic. It accelerated digital transformation and the adoption of emerging technologies like artificial intelligence at an incredible pace. With the prominence of the remote-work model, companies, both big and small are implementing artificial intelligence for various use cases including automation of day-to-day tasks to stay operational and profitable. In a world undergoing intense digitization, AI is no doubt opening up a plethora of new opportunities and scenarios. To understand the implication of AI in 2021, we asked 100+ industry leaders around the globe for their top AI predictions and insights.
100+ Industry leaders share their AI predictions and insights for 2021
In this article, we have handpicked the top predictions on how artificial intelligence will be leveraged in business, the likely challenges to AI adoption, and tips for implementing this emerging technology in 2021.
The rise of human-centered AI and automation at scale
The pandemic forced every organization to automate complex or manual processes to free up employee time and focus on customer success. Because of the pandemic, this technology is here to stay in 2021 and beyond because there will be a true need for automation that helps humans to do their best work. Roughly 70% of business leaders spend about one to three hours per day on mundane tasks, so if you removed the tedious projects from their plates, it would free up an entire week and a half per month.
There are some things computers can never do — humans have the capacity for creative thought unparalleled by artificial intelligence — but computers are great at removing the repetitive grunt work from our days. That’s the beauty of AI: it allows us to thrive in our everyday work by using strengths that are distinctly human.
The COVID-19 pandemic also amplified the need for and use of bots. Businesses and governments rapidly adopted bots to scale to unprecedented demand as customers turned to online channels amid in-person closures. They’re widely being deployed in customer service apps, and being used for sales, marketing, and e-commerce as well. They’re even being used for sensitive inquiries in real-time on topics such as where to find a testing center and how to file for unemployment insurance. I expect this trend to continue in 2021 as consumers around the world become accustomed to the use of bots. Remember email? That was a pretty foreign concept in the 1970s. Now can you imagine your life without it?
AI will become part of everything and you’ll hardly notice -By the time we hit 2022, AI will finally have entered the mainstream. The pandemic introduced countless new digital touchpoints for B2C and B2B companies alike, which means there’s more data than ever before. Businesses and consumers will have more of an understanding of what AI can do to reduce human errors, create more intelligent experiences, and generally make people’s lives easier. This is where AI shines, and why it is here to stay as the unifying force behind all key technology advances serving employees and customers. The data tells a similar story as IDC predicts that global spending on AI will double in the next four years — reaching $110 billion in 2024 — as companies see an opportunity to boost innovation, improve customer service and automate routine tasks so their employees can focus on more strategic work.
Autonomous optimization and AIOps will be the unsung heroes of the remote-workforce
In 2021, autonomous optimization and AIOps will be the unsung heroes of the remote-workforce. Already stretched thin and facing cost pressures, applications and IT teams will be increasingly tasked with monitoring their infrastructure to ensure overload is not taking place amidst the remote application and VDI usage. Traditional tools have usually been rule-based and inflexible, which has made it difficult to monitor, optimize and troubleshoot hybrid cloud environments during an enormous growth in data and the unexpected challenges of COVID-19. To this, we’ll see an increase in application owners preferring infrastructure that continuously optimizes the workloads, with IT teams implementing AIOps — using artificial intelligence and predictive analytics to gain essential insights into their infrastructure status — allowing them to focus on innovation work and alleviate them of network or data monitoring bottlenecks.
Now, more than ever, teams are relying on AI technology to help digest and analyze critical data to help inform operations across the business — especially during a time when increasing efficiencies and cutting unnecessary costs is on top of mind to ensure longevity. In 2021, we will continue to see industries adopt AI solutions, but many will find they do not have the infrastructure in place to handle the data needed to feed the AI/ML pipelines at the speed and scale required. This will especially be true post-COVID, as companies look for more ROI out of their digital transformations. The result will be a boom in software to improve productivity and quality of life for IT and DevOps teams.
2021 will be the year of AI democratizing and driving software adoption at scale
The trying times of 2020 have shifted emphasis on how we view and perform work. The new normal of enabling work from anywhere is heavily dependent on the accessibility and adaptability of digital tools. Enterprise software adoption however suffers due to their inherent complex nature and by extension the workflows that run on them.
2021 will be the year of AI democratizing and driving software adoption at scale. AI-powered algorithms and logic are going to play an increasing role in smoothening the software experience across the lifecycle — from onboarding, support, change management, and beyond. AI will help with deciphering user intent, predicting next actions, and fine-tuning interactivity to raise the bar of personalized experiences thus driving the adoption of critical software, at scale.
Enterprise applications will find AI augmentation either through out of the box capabilitiAI driven es or through stand-alone software like Digital Adoption Platforms and will witness enhanced adoption across Marketing, Sales, HR, Supply Chain, Finance, and other functions.
AI Cybersecurity tools and Quantum tools will become the differentiator
AI Cybersecurity tools and Quantum tools will become the differentiator between the weak and strong companies, much the same way that stone castles differentiated strength from mud huts in the early centuries. AI and Knowledge graphs will start to become the foundation for understanding disparate enterprise data, much the same way that relational databases are essential for financial transactions. Advanced degrees that facilitate AI will become the new MBA. The new demand for AI data scientists to enable business results will continue to increase the need for master's and PhD graduates with data science and AI expertise.
One of the largest problems with AI is that it is often left to a small group of AI data science that is working on AI science projects on a shoestring budget looking for academic level results, rather than AI being driven by business unit executives as a means to an end to make real tangible business outcomes. I see that over the coming year the gap between those that experiment with AI with 2 to 5 data scientists playing with the latest tech toys and those that extract hundreds of millions of dollars from AI-driven by business executives will widen. Companies that are looking to generate value from AI need to;
1) Make business unit executives accountable for results from AI including using AI to better segment customer markets, using AI to improve customer service experience, and using AI to reduce costs by automated manual processes.
2) Invest in the right AI resources, if you can’t attract those skills to your company then find a good partner firm that can provide you the skills and teach you to develop this competency and
3) Invest in the right cross-functional business team to meet specific business objectives through the use of AI.
Endpoint devices will become increasingly smarter
2021 will be the year that the industry realizes that IoT is merely a vehicle for data. It’s how organizations utilize the data to generate new revenue streams or increase customer satisfaction that is the true driving force behind connecting billions of assets.
Endpoint devices will become increasingly smarter. Through machine learning and artificial intelligence, network intelligence is moving closer to the edge, giving endpoint devices new roles and tasks that make them smarter. With compute power delivered by low-cost hardware, powerful intelligence can now be achieved through machine learning algorithms that can live (train and process) directly at the data’s originating source. In addition, hardware capabilities are ever-evolving along with the software living on the hardware. The industrial sector is already benefiting from the efficiency gains of processing data locally to understand what’s happening to these machines and devices. Other markets need to shift and adapt to local data processing or specifically machine learning on the endpoint as it will be essential in helping organizations scale.
Cloud data transmission costs will increase dramatically. The cost of transmitting, storing, and processing data in cloud environments is a growing pain point for organizations of all sizes. In 2021, Edge AI will mature from innovation projects to an industry standard. Bringing edge and embedded AI to the endpoint for processing data locally eliminates the cost pain point. The bottom line: If businesses aren’t doing innovative things to derive more value generated from their data to keep pace with expenditure, they’re just going backward.
Predictive maintenance will top manufacturing pain points in 2021. Semiconductors and chips use simple decision trees that can grow to become complex. Shifting from a decision tree-based process to a predictive algorithm is the next step for intelligence at the edge. Manufacturers are in the business of producing items. They are not in the business of technology and data management. As data grows in volume, there is an ever-increasing need to make sense of the data and act upon the insights to remain as efficient and productive as possible. To do that, manufacturers need to rely on technology to help them realize the value of the data they are generating. This will come from deploying local/Edge AI for processing data locally, for producing insight-rich output. The outputs from Edge AI solutions will ultimately automate response and action to allow for less dependency on human interaction and more so on automated corrective action.
Customers always come first, regardless of technology
Customers always come first, regardless of technology. Forward-thinking companies are implementing robotic process automation (RPA) and artificial intelligence to reduce or eliminate repetitive, time-consuming tasks, so their employees can better focus on customer engagement, customer satisfaction, and other valuable contributions. It’s counter-intuitive that automation can allow companies to offer more 1-on-1 interaction and personalization, but it’s true! We especially see opportunities with judgment-based processes involving unstructured (non-digitized) data in the banking, financial services, and insurance sector. Some examples include fighting financial crime like anti-money laundering measures or keeping constant watch for negative news reports. We expect this trend of “improving customer satisfaction through AI-driven RPA” to amplify considerably as we move through 2021.
We see that both executives and rank-and-file employees are often eager to automate tedious, time-consuming, and costly manual tasks. But sometimes middle management is more hesitant because their identity and success metrics rely on how many people they supervise. Champions of automation can overcome this roadblock by steering their organizations toward fundamentally changing their definitions of management and success — focusing on overall productivity/capacity and increased employee satisfaction, instead of someone’s number of direct reports. Also, automation may seem like a big project, and companies may feel they don’t know where to start. We suggest starting with pre-built use cases that have already proven to be effective for your industry. Ultimately, AI and automation inevitably lead to increased productivity and a better employee experience, so however you begin, it’s going to be a win.
As with everything, people are essential to the success of businesses’ AI strategies. Forget the tired argument that automation is killing or stealing jobs — we see the opposite. Rather than cut valuable talent, companies create or update employee roles as automation is integrated into operations. And employees who are freed to direct their efforts to higher-value tasks and more personal connections with customers are easier to retain and elevate toward increased productivity and success for both themselves and the whole company.
Customer service conversational interfaces are becoming exponentially smarter
One of the biggest areas I’m excited about in 2021 is the rise of Natural Language Understanding (NLU). This modern branch of Natural Language Processing (NLP) uses Deep Learning to solve human-level cognitive tasks on language, like translation, question answering, and logical inference from text. There’s been a kind of NLU renaissance over the past 3–4 years, culminating in models like BERT, which has achieved state-of-the-art accuracy on tasks like translation, question answering, and logical inference; and GPT-3, the model out of OpenAI that can literally generate Shakespeare
And these technologies are going to have wide-ranging impacts on enterprises and consumer industries. Customer service conversational interfaces are becoming exponentially smarter, and AI can be used in recruiting to match resumes to jobs or to detect misinformation in the news. Natural language is everywhere, so I’m excited to see that the technology is finally catching up to be able to understand it as humans can.
AI industry finally takes bias in artificial intelligence seriously
2021 will be the year the AI industry finally takes bias in artificial intelligence seriously. In 2020, the Covid-19 crisis hit certain communities far worse than others and helped fuel tensions over biases embedded in our societies, which led to a groundswell of protests such as in the Black Lives Matter movement. AI models cannot live in a bubble and they cannot continue to accentuate the biases prevalent in society. I predict we will see greater pressure for the AI industry to improve the reduction of bias. We should also see a greater effort towards more systematic research to find solutions to these problems.
AI is created by humans and humans will always have a bias. The first step to reducing bias in AI is to increase diversity in the teams that build it. Having different views on different topics will surely help build better systems. AI bias tends to amplify problems that already exist in any given system.
To help solve the AI bias problem, in 2021 we are likely to see far more vendors acknowledging the need to be transparent about how they devised their models and what steps they took to counter bias through diversity among creators and testers, as well as datasets stress-tested for bias.
Also, the industry will increasingly find more tools that are easy to use without training. That will be a great help because it will make the solutions available to a wider audience which can then provide a feedback loop to help reduce biases.
The industry will move towards high efficiency AI to satisfy market pressure
The cooperation of people involved in business processes is massively impacted by the pandemic. As the local information resources in offices are not available anymore, a digital method must replace the direct information exchange. As a large part of communication has moved to emails and website requests, a growing number of intelligent filtering and routing tools will help agents in the home office process the flow of incoming messages and documents. Because of the complexity and speed of the data involved, organizations need to incorporate AI in their systems in order to improve the efficiency of business processes. This will affect all industry verticals. And companies will look for practical and useful AI solutions, not the ones that achieve high precision under experimental conditions, but those who are production-ready and bringing them immediate value. In 2021, the AI industry will move towards high efficiency AI to satisfy market pressure.
To me, the lack of efficiency of current models is the main impediment for large-scale AI adoption. It turns out that in practice only a highly efficient AI system can be tuned to a useful level of precision. The second roadblock I see comes from the time and effort required from subject matter experts, who are asked to meticulously transfer their skills to an automated system that will potentially replace them — at least this is how they can see it, if they have not been involved in the project since the beginning. Nobody likes to reveal the professional tricks that make him successful, especially not if one fears for one’s own position within an organization. Finally, the fact that many AI projects are introduced in a bottom-up fashion — suggested by practitioners at the operating level to the upper management — constitutes a major impediment for the deployment of AI projects in enterprises.
For a large-scale adoption of AI, there needs to be a top-down strategy in the first place. The C-level must be convinced of the added value of the AI project and be able to convince all other levels that they will benefit from it. In particular, subject matter experts should be involved in the AI adoption process right from the beginning (in the planning phase) so that they do not fear for their job, but rather become the owner (master) of the AI. Once the decision to roll out an AI project has been made, the management should seek to motivate practitioners by providing them with adequate time and budget resources so that they can play around with AI and gain the benefits needed.
Customer Service via transformer AI’s is also going to have a massive uplift
I believe that all kinds of businesses would adopt different types of customer experiences that are centered around AI applications. Because the telecom, media, and entertainment industries have massively grown in the adoption of technological advancements, it is only going in an upward direction and is going to be complemented by big data and cloud-native technologies.
With the innovation of AI integrations with communications over the past year, technologies like 5G networks, IoT, AR/VR, the telecom and media industries are just at the start line of such advancements integrated with AI and are increasing in popularity. The implementation of 5G throughout North America, Europe, and East Asia will certainly increase by 2021 and allow for almost seamless network coverage, and advanced traffic management.
Wireless media and connectivity are also on track to becoming mainstream, where volumes of data are easier to process through on-device processing or cloud over low-latency networks. Customer Service via transformer AI’s is also going to have a massive uplift. This will be done with the help of intelligent bots in the NLP domain, automating translations, emotional analysis, and search question contextualizations. This will prove more effective with omnichannel cloud communication which includes but isn’t limited to voice and video messaging.
Author: Tharika Tellicherry
Originally published at https://www.crunchmetrics.ai on January 20, 2021.