How CIOs Can Anticipate Challenges When Adopting AI
Share this on:
It’s no secret that the adoption of AI technology is advancing at a rapid pace. It doesn’t matter if you’re making cars in the US or providing business phone systems in Canada, AI is a growing part of the global business landscape.
In fact, according to PWC’s recent AI Business Survey, 56% of respondents reported widespread adoption of AI in their business.
Image sourced from PWC
Why do companies adopt AI?
Businesses in the above PWC survey stated they were twice as likely to report substantial value returns from their AI initiatives. With that in mind, you can see why the widespread use of AI is becoming more popular.
According to the same study, these successful adopters are using AI across their departments to drive three main business outcomes:
Proceeding with these three goals at once yields more rewards than taking a more reserved ‘one-step-at-a-time’ approach.
As for the more direct question of “what do these businesses use AI for?”, there are AI applications in almost all business areas. The main uses identified by PWC’s ‘AI leaders’ are in:
But that’s still not an exhaustive list of AI’s wider uses. The bottom line is that businesses are using AI to improve processes, decision-making, and systems. On some scale, AI adoption is happening across all industries, though some have been quicker to adapt than others.
What industries are adopting AI?
It might not come as a surprise to learn that the tech and computing industries have been the fastest adopters of AI. According to this O’Reilly survey, the next biggest users are financial services, followed by healthcare, education, and public services.
Telecoms, manufacturing, and retail are all seeing benefits from AI adoption, too, though widespread adoption across those industries is slower. Other adopters included energy suppliers, and the entertainment and security sectors. That’s a lot of potential AI use cases.
The Main Challenges of AI Adoption
AI isn’t a catch-all solution for businesses. There are different types of AI programs with different applications. For example, running an auto attendant for a hosted PBX in Canada will have different data and upkeep needs than an AI analytics program for your production line.
There are two main issues with data, usually encountered at the AI training stage: the availability of data and the quality of data. Machine learning, in particular, needs large amounts of data to train its algorithms.
Sometimes, you may have an abundance of data but it’s not the right data type for the AI to make good use of. For example, let’s say you’re using AI to identify errors in an automated process. You have a lot of data from the process functioning as normal, but that might not be helpful.
Because the errors are an uncommon occurrence, there’s less data. So, you can show the AI what correct functions look like, but have insufficient data on the error types for it to accurately identify them in production.
Training people to work with AI-assisted tools is relatively simple. In the higher skill tiers of AI development, though, there’s a definite skill gap. That means that businesses often have the capability to implement AI applications, but not the skills to adapt them or fix problems.
Much of the current skill shortage is to do with AI being a relatively new technology. Workers with extensive experience working with AI are yet to emerge in larger numbers. This can be solved in time as long as AI-focused businesses invest in employee development.
Integrating AI-assisted technology into your existing business functions can be a significant hurdle. This can be especially challenging if you have legacy infrastructure or a business culture that has been reluctant to embrace new technology.
Analyzing where AI can assist your current functions most effectively is vital before proceeding with implementation. Work with your departments to identify potential efficiency savings and more complex processes that need human oversight.
Investment cost & ROI
Proving ROI with a long-term AI deployment can be difficult. This is one area where CIOs often experience the most pushback on AI initiatives. The internal skill gap means that third-party consultation is often needed, and technology costs for AI development can be steep.
Some of this problem is already being eased by a wider availability of third-party vendors. Plus, the benefits of cloud-based computing have made AI-assisted processes available on an on-demand basis.
Security & transparency
Data security and privacy have been big concerns for both tech companies and regulators. Huge amounts of data, sometimes personal, are being accessed and interpreted by AI algorithms. So, it’s only natural that concerns over transparency have been raised.
IT security professionals have suggested that AI systems don’t provide any greater risk to data security than other online systems. Potentially, the greatest risk is a lack of understanding of data security from those implementing AI applications.
Best practices for CIOs adopting AI
Despite the hurdles, companies are finding ways to return value on AI applications. CIOs that work for startups can even leverage AI for small businesses if they are familiar with these best practices for AI adoption.
AI is often referred to as one blanket term, but there are actually different types of AI applications. The most popular ones currently in use are machine learning (ML), natural language processing (NLP), computer vision, AI analytics, and explainable AI (XAI).
These AI types have different functions, like NLP for sentiment analysis or computer vision for use in controlled environments. Getting to know the types of AI that are relevant to your industry is important for modern COIs.
Know the limitations of your AI
The other side of knowing your AI comes from understanding its limitations. Naturally, applying an AI that’s designed as a chatbot to sort accounting data isn’t going to yield great results.
That might seem obvious, but even functions that seem highly related to us can have important differences to an AI. Knowing what your AI can’t do is as important as knowing what it can do.
Maximizing data & data scientists
Before AI came along, businesses relied on data scientists for analytic insights. That doesn’t mean that AI makes data science obsolete though. In fact, data scientists can be highly beneficial to successful AI implementation and use.
That knowledge of data and analytics means that your data science team is the most knowledgeable about the data your AI uses and how it interprets datasets. They can assist you in mapping the data pipelines you’ll need for a functional AI.
AI and your business
From healthcare to telecoms, AI-driven applications are being used to modernize business processes across various industries. Preparing your company with the knowledge and skills for AI adoption is essential for future-proofing.
About the Writer
Jenna Bunnell is the Senior Manager for Content Marketing at Dialpad, AI-incorporated cloud-hosted unified communications system that provides valuable call details for business owners and sales representatives through products like Dialpad’s Canadian VoIP business phone. She is driven and passionate about communicating a brand’s design sensibility and visualizing how content can be presented in creative and comprehensive ways.
Disclaimer: The author is completely responsible for the content of this article. The opinions expressed are their own and do not represent IEEE’s position nor that of the Computer Society nor its Leadership.