In 2022 AI innovation will continue going stronger than ever before, and at the same time, remain the most transformative technology we have developed as humans. This comes as no surprise as even today, we use some kind of artificial intelligence in nearly every aspect of our lives, from website chats helping us with customer requests, to solving some of the greatest social, economic and environmental challenges, and entrusting critical business decisions to it. AI promises unprecedented, sweeping solutions to companies looking for ways to adapt to the COVID-19 consequences and accelerated digitalisation.
Challenges and factors driving AI trends in 2022
However, organisations are facing some big challenges with AI, acting as factors shaping the AI trend we’ll see in 2022.
In 2020, companies spent $50 billion on artificial intelligence projects globally. And still, only 11% of AI adopters saw a sizeable return on their AI investments. Enterprises are struggling to implement AI at scale. The biggest reason for these somewhat discouraging results is a failure to get a buy-in from the C-Suite, and not being able to link the technology capability part with the business.
Another figure is that almost 80% of all AI/ML projects are stalled between proof of concept (POC) and full-scale deployment. The main reason behind it is that launching pilots has proven deceptively easy, but deploying them into production is notoriously challenging. This is the AI pilot paradox that data and AI practitioners are discovering the hard way. “Although the potential for success is enormous, delivering business impact from AI initiatives takes much longer than anticipated,” says Chirag Dekate, Senior Director Analyst at Gartner.
Consequently, as the AI investments keep on failing to provide the ROI, the cost for training models will get higher. The frustration from not getting the expected results will become louder. Companies may inevitably start pausing their investments and abandoning the initiatives altogether.
Trends we can expect to see in 2022
We have been listening to the market signs and trends experts predict we’ll see in 2022 that will provide the solutions needed to help organisations accelerate AI deployment at scale.
ROI-driven AI implementation
The challenges with AI above show that we need a different approach with AI investments and deployment. The failure to realize financial outcomes from their AI investments is the reason why one of the biggest 2022 AI trends will be an incremental, ROI-driven approach to AI development, states ITRex.
ROI-driven AI engineering and implementation will help organizations:
- Consider the business outcomes of deploying AI innovations beforehand
- Foster collaboration between business and IT teams
- Ensure C-Suite support
- Focus on a limited number of use cases for an AI pilot while devising a broader implementation plan
- Upskill existing IT teams and consult with experienced artificial intelligence companies
- Discard vast piles of historical data, which has become less relevant since the COVID-19 outbreak, and take into consideration wide and small data
- Learn from failures, retrain algorithms on new data, and create a continuous loop for machine learning model redeployment.
Accelerating enterprise use of AI
By 2025, companies that have adopted AI will be 10 times more efficient and have twice the market share of companies that have not, says Vikram Mahidhar, Global Business Leader, AI solutions, at Genpact. 2022 will come with greater opportunities for organisations to introduce a new approach that will allow them to realise the benefits of AI faster than ever before.
For companies to accelerate the use of AI, they can no longer afford for AI to be on the fringes of their organisation. It needs to be at its core – as the neural wiring, stitching the organisation together, in order words AI must be a fast-tracked priority for every enterprise, emphasises Genpact.
With old, legacy systems, and traditional data warehouses, accelerating AI can look like an impossible undertaking. The changing requirements of AI require organisations to have a hybrid setup in place grounded in an open-sourced AI platform take should have the following features:
- Simplicity: The system should be simple enough to integrate efficient software that works. The AI platform should be easy to develop and implement for starters.
- Ease of use: The data should be easy to use for the AI platform. The right AI platform will clear the environment and make way for easy data storage and working methods.
- Faster model: An AI server should have faster training times: Machine learning and deep learning shouldn’t take as much time as they currently take.
- Open AI platform: A platform that partners can build upon. This program will make the shift to AI easier for many organisations.
For enterprises to truly accelerate the use of AI, they should abandon the traditional thinking and approach, and adopt the right mix of tools and technologies apt to help them deploy AI at scale.
Next year, we will also see a rise in smart, AI-native companies using modern cloud-based technologies, like AI and the Internet of Things (IoT), to better use its data to get real-time insights about its operations, market environment and customers.
These intelligent enterprises are not giant Fortune 500 companies. On the contrary, it’s the SMBs that are some of the most innovative businesses that have managed to transform their business models to become more customer-focused, more flexible and more responsive to market demand. The businesses of the new era are leveraging AI to improve business processes and automating previously tedious back-office finance operations like receivables, managing global workforces, research, designing new products while forecasting their revenues and many other activities.
“The Intelligent Enterprise is not a product. It’s a mindset of using modern strategies and processes, powered by an intelligent suite of technology solutions,” says Steve Tzikakis, SAP President South Europe, Middle East and Africa. It’s an approach that enables companies to tap into all available human and machine-generated information to win in a highly competitive economy, by differentiating on the insights they can turn into action. Intelligent enterprise creates a transformative effect on their workforce, as technology frees employees from mundane tasks and allows them to focus on the customers, transforming both the Employee Experience and the Customer Experience with the power of data and AI.
We started talking about Data Mesh a lot in 2021. In short, it presents an emerging paradigm shift in data architecture that is moving away from centralised data lakes and databases, and going towards data domains and data products, changing the way we think about data as a by-product of production processes, and start thinking about data products as first-class citizens, while ensuring that the ownership and responsibility of such data products lie with those that know the data best.
In 2022, Data Mesh will become the new fabric for distributed data, confirms Qlik. The need for faster access to data across increasingly distributed landscapes is driving integrated data management that uses metadata, semantics, real-time and event-driven data movement, and orchestration in the pipeline. Putting these capabilities into a distributed architecture is being referred to as a “data fabric”, which leads us to the next trend to expect in 2022.
Data fabric is defined as integrating data across platforms and business users. The idea is to make it easy to use the data you have and reduce data management efforts. As the number of data and siloed applications soared in the last decade, data fabrics have emerged as the solution to simplify an organisation’s data integration infrastructure and create a scalable architecture that reduces the technical debt seen in most D&A teams due to the rising integration challenges.
Data fabric addresses the challenges of data fragmentation and data residing in disparate systems by providing an orchestrated approach for collecting, unifying, and governing data sources throughout the enterprise data management system.
In 2022, we’ll see CTOs become increasingly interested in data fabric as they grasp the benefits of a unified architecture that helps organisations manage and maximise the value of their data.
Another trend related to data fabric with an added layer of automation is the automation fabric. Forrester’s Automation is the New Fabric for Digital Business report finds that the collection of automation technologies, including RPA, low-code tools, chatbots and machine learning, are converging on top of the application layer, coevolving into a broad weave—a “fabric”. This automation fabric is where we can expect to see digital business happening in 2022.
As low-code tools and robotic process automation builders become more available, we can expect to see an expansion of automation fabric. It combines digital workers and AI agents such as chatbots with event-based and integration-centric orchestration.
Automation fabric, however, is not a product available on the market. Organizations can buy the technologies they need to flexibly address their transformation goals, and create a system for whole-of-business automation to support the goals of human-centred automation and an autonomous enterprise.
Low-code and no-code AI
In 2022, access to artificial intelligence and fast deployment of applications will become even more critical with the inflating popularity of cloud-based software development. The low-code application development market is expected to increase from $4.32 billion currently to $27.23 billion by 2022, according to ResearchAndMarkets. Gartner also predicts that 65% of all apps will be developed on low-code platforms by 2024, by which point Gartner also predicts that 66% of big companies will be using low-code software.
Additionally, with the widening of the data scientist talent shortage, companies will increasingly turn to low-code or no-code tools. These platforms allow tech and business professionals with no coding experience to build apps and potentially fill talent gaps in their organisation.
Low-code and no-code AI tools reduce the time it takes to create software, through graphical user interfaces, instead of traditional computer programming and replace building apps visually by dragging and dropping UI components. Thus, business apps can be delivered more quickly, and a wider range of people (sometimes called “citizen developers”) in an organisation can contribute to app development.
Bringing AI into the hands of the business users
More and more, we see tech giants like IBM and Amazon creating low-code AI platforms that enable users without a coding background to more easily access the benefits of AI. With the expansion of low-code and automation builders, in 2022, AI will become even more accessible for innovation to the business users who will be able to build their own workflows and customised functionality.
Previously, AI technologies were limited to highly technical experts who didn’t have domain expertise. But for the business to truly benefit from AI, business people need to be able to have first-hand access to AI tools. Using low code for AI allows citizen developers and data scientists to utilise AI building blocks to generate an AI engine that fits their needs. “It puts ‘AI superpowers’ in the hands of users, eliminating the need to write complex code, compile, deploy, and scale it,” says Kfir Yeshayahu, the SVP of products at Veritone for Datanami.
The pandemic and need for remote work has had a deep impact on AI innovation. With the advancements in artificial intelligence and IoT technologies, organizations are getting closer to AI-augmented automation, which will be crucial for automating business processes and enhancing productivity and workflow by combining humans and AI.
Here are some stats that confirm that AI-augmented automation will be a leading AI trend in 2022 and beyond:
- 80% of executives are currently accelerating their business process automation efforts, states the World Economic Forum.
- 25% of companies already use AI in workflow automation, while 51% of enterprises are planning to do so shortly, reports Salesforce.
- By 2023, 40% of infrastructure and operations (I&O) teams will use AI-augmented automation in large enterprises, freeing up IT personnel’s time for strategic work, says Gartner.
Rise of responsible AI
With the accelerated adoption of AI in 2022, responsible AI will take a front and center place in the AI field. As countries implement national AI strategies, they will need to seriously consider the moral implications of AI, ensuring positive automation outcomes for all parties involved. Enterprises as well will need to create AI systems that explain the rationale behind their decisions and do not discriminate against ethnicity, age, gender, religion, or place of residence. By 2023, all AI professionals will have to demonstrate a good grasp of responsible AI principles to secure a job, emphasises Gartner.
Moreover, Forrester predicts that the market for responsible AI solutions will double in 2022, as some regulated industries have started adopting responsible AI solutions that help companies turn AI principles such as fairness and transparency into consistent practices.
Next year, the demand for these solutions extend to other industries and verticals deploying AI in their business. With the increase in adoption of responsible AI solutions, existing machine-learning vendors will acquire specialized responsible AI vendors for bias detection, interpretability, and model lineage capabilities, Forrester concludes.
Next year will bring many new, exciting innovations in data and AI, inspired by organisations’ lessons learnt in the AI pioneering era. Until now, most organisations have explored strategies, technologies, have one or several AI pilots deployed in production – and have gained great value from it.
Now, in the AI revolution decade, it’s time to finetune the innovation engine and make it available to the business so they can start creating value with it. Henrik Göthberg, the Chairman of the summit, calls it “The rise of the trinity: Data Engineering, AI engineering, Value Engineering”.