According to a recent McKinsey global survey, the adoption of at least one AI capability has progressed to nearly half of the global enterprises. At least 20 percent of the respondents to the survey reported having adopted at least one of the nine AI in their organizations.
Telecom, financial services, and technology sectors are leading in AI adoption, with retail following close behind. The picture becomes a bit clearer if we take a look at this AI adoption graph, by industry.
<Refer image under “Where AI is being used” section:
Fundamental barriers remain with companies fumbling with the correct AI adoption strategies. AI adoption is tricky, and many companies fail at implementing it correctly within their processes.
Key challenges companies face while adopting AI
- The skill-set required to plan for an AI solution is pretty specific, and every organization may not have it at hand
- AI technologies are generally on the expensive end for most organizations
- Careful planning needs to be done to figure out the quality and kind of data that is to be fed to your AI solution because eventually, it will be this data from which your solution will learn
- There are a ton of ethical challenges to be dealt with when thinking of adopting AI such as the opposition of bots mimicking human conversations.
Common pitfalls to avoid when implementing AI
Gartner’s Magic Quadrant for the CRM Customer Engagement Center predicts that up to 40% of the AI bots / VA applications launched in 2018 will be abandoned by 2020.
What can prevent you from being one of that 40%?
- Bad start – Companies end up spending on use cases with little value (lack of business case)
resource intensive. Avoid critical or High-risk use-cases. Instead, you may want to consider a medium risk. Go for a quick win.
Investing in AI is a big decision for most enterprises; on top of that, there’s the expectation of witnessing significant ROI from it within the shortest span of time. Thus, it is imperative to select the right business use case to optimize with AI.
- The missing link – ‘AI-ready’ dataset
Being the foundation of AI, you have to have a clean dataset. It takes enormous time to collect, clean and organizes it. Ensure the data suffice to build a deep learning based model. Overzealous executives tend to force AI down the management without really having adequate data for AI/ML setup. The best strategy would be to put together a task team to make the Data ‘AI-ready’.
Technology has changed at a dizzying pace. Ensure that you’re always up-to-date with your understanding of AI technology and its latest advancements be its voice recognition, virtual agents or so. A good thing to consider would be whether your proposed AI solution will integrate well with your existing tech structure. You will also need a robust Cloud Infrastructure and Data Warehousing. AI technology is generally on the expensive end for most organizations.
The skill-sets required to build any AI application are very specific and many organizations may not have it hand. Identify and deploy people with the right AI skill-set at any particular stage of AI adoption. With new breakthrough and papers appearing every day in AI, the team should be research minded and application-driven both to cope up.
Is implementing AI worth the time and money? A PoC would be a wise decision, especially in the case of AI
- There are many vendors who claim they understand AI, but do they understand you?
- Have you chosen the right use case? Do you have a ‘clean’ dataset?
- Will it integrate with your existing tech structure?
- Did you cover all corners when estimating the effort and cost?
How ready-to-use AI platforms are accelerating AI adoption
AI automated platforms like H2O’s Driverless AI, Amazon’s SageMaker, Hotify’s Nugene, and others lend a helping hand at making some of the stages of AI adoption easier and more risk-free. Such platforms combine the core elements of AI such as machine learning, automation, and natural language processing, with cloud infrastructure to bring an AI-ready deployable platform. Additionally, investing in research and building a talented team will help the most in your long term AI strategy.