The year 2020 will be remembered for many things. It was the COVID year, a year of social change, and a year of significant political shift in the US. In the tech world, 2020 will be looked at as the year AI went from being the stuff of science fiction to a business reality.
Organizations of all sizes accelerated their artificial intelligence (AI) initiatives when COVID-19 hit in 2020. Now more than a year into the pandemic, organizations are focused on understanding their data and using AI to support internal processes. Years ago, there was an axiom that stated “all companies are technology companies” but that narrative has been played out. Today all companies are data companies, but the massive quantities of information require greater automation to find key insights. Enter AI as the solution to this growing problem. AI is reaching broader market maturity, with budgets growing more than 50 % year-over-year, ranging from $500,000 to $5 million annually.
These findings come from the 2021 State of AI and Machine Learning Report released by Appen Limited, a provider of training data used to build AI systems. Appen, in partnership with Harris Poll, surveyed 501 professionals online in March 2021. The respondents consisted of 251 business leaders and 250 data scientists, data engineers, and developers at US-based companies with more than 100 employees.
AI success requires a significant budget commitment
The data shows AI budgets of $1 million or more had a higher probability of AI project deployment than budgets under $1 million. Additionally, budget allocation correlated with achieving a return on investment (ROI) from AI deployments. Forty-eight percent of organizations with budgets of $1 million to $3 million and 49% of organizations with budgets of more than $3 million reported a deployment rate of 61% to 90%. The numbers were significantly lower for organizations with budgets under $1 million.
Budgets also have a significant impact on market leadership perception. Appen broke down organizations into several categories: those spending less than $500,000, those spending $500,000 to $1 million, those spending $1 million to $3 million, and those spending over $3 million. Organizations spending less than $500,000 on AI projects were less likely to consider themselves market leaders compared to the high spenders.
AI is now being deployed as an IT tool
Another significant finding uncovered in the report shows responsibility for AI projects is shifting from business decision makers to technologists. For 39% of organizations, C-level executives were responsible for AI initiatives in 2021, a sharp decrease from 71% in 2020. A rising number of technologists implementing and maintaining projects is an indication that AI is progressing within organizations.
Many organizations are moving away from the belief that AI is a “silver bullet,” and instead, they’re incorporating AI into IT operations. When asked how organizations are employing AI, 62% of the respondents said they’re using it to support internal IT operations, 55% use it to better understand corporate data, 54% to improve productivity and efficiency, and 49% to aid in research and development of new products. Reducing business costs was also at the top of the list for 45% of the respondents.
AI success comes from high-quality data
A successful AI deployment always depends on obtaining high-quality data, which is a struggle for organizations both large and small. For this reason, many are turning to external data providers to help with data acquisition, preparation, and management. Organizations with an external data provider were twice as likely to fast-track their AI initiatives compared to those that didn’t have one. In fact, the majority of organizations surveyed have partnered with external training data providers to deploy AI projects at scale.
Working with a data provider not only helps organizations deploy more AI projects, but it also produces meaningful ROI. According to the respondents who use an external data provider, 42% reported a positive ROI for 61% to 90% of their AI projects, compared to only 32% of those deploying without a provider. Companies that use external data providers were also 1.8 times more likely to say their company was able to address AI-related privacy and security issues.
Ethics and interpretability top AI concerns
Despite organizations embracing AI in the wake of the pandemic, the report found disagreements that exist between business leaders and technologists, mostly over ethics and interpretability. Technologists are more concerned about ethics (41%) than business leaders (33%), while business leaders care more about interpretability (47%) than technologists (38%). Nevertheless, business leaders and technologists agree responsible AI projects can be carried out with a risk management approach.
All the respondents confirmed that their organizations (regardless of size) will continue to accelerate the development of AI post-COVID-19. That’s why organizations are increasingly leveraging training data providers and updating models on an ongoing basis. More specifically, 87% of organizations are updating their models regularly and 91% of large organizations are updating at least once per quarter. Organizations that do this are able to maintain data accuracy and feel they are ahead of the AI game.
It’s my belief that AI has the power to be the most transformative technology since the Internet. Companies have to be careful to ease into deployments and have AI solve some big problems first, such as improving IT operations, R&D, and even some aspects of customer service. Once organizations have gained experience with AI and understand the pitfalls and learn some best practices, they should be aggressive about rolling out AI more broadly.