Artificial intelligence plays a big role in today’s technology-driven world. The high demand for artificial intelligence (AI) is fuelled by the benefits it has enabled for organizations. Netflix gained significant growth in subscriber base through AI-driven personalized movie recommendations to customers¹. Amazon developed an AI-powered feature that gives recommendations, considering factors like brand, price range, and customer reviews after the customer uploads a photo or screenshot of the desired look². After looking at such success stories, almost every enterprise has been trying to incorporate AI, more often ending up force fitting it.

According to a recent research by IDC, half of the AI projects fail in one out of four companies3. For example, Amazon had built an AI-based recruiting tool that ended up being gender-biased, favoring male applicants over female applicants. After such failures, organizations realize success cannot be guaranteed despite big investment son time, money and effort.

Why do AI efforts fail?

When we analyse what resulted in AI failures, one main reason is failing to understand the capabilities of AI technology. In fact, as per the IDC research, the two leading reasons for the failure of AI projects were lack of required skills and unrealistic expectations³. Other reasons for failure of AI implementation are insufficient research, improper analysis of the problem, insufficient planning or prioritization of tasks, etc. One important takeaway of the analysis of the issues is that AI is still in its nascent stages and may not be the best at solving all kinds of problems.

Getting AI right

So, what is the right way to solve such problems better? ‘Seven Habits of Highly Effective People’ by Stephen Covey, puts forward the thought- begin with the end. This emphasizes on having a clear picture of the ultimate goal so that every single step taken ensures you are on the right path. The following points can help in increasing the success of an AI project:

1. Assess your problem– Understand the problem and analyse whether the problem can be solved by application of AI technology. Do you have relevant and sufficient data to solve the problem? How will the result look like? Who will be the end users? Will the solution be compatible with all the end users? Do you have competent people in analytics and AI?
2. Understand the impact of project - Determination of the value or business impact that the project solution will deliver is paramount. The higher the impact the better. The value of AI solutions need not be limited to ROI: it may also include more secure processes, better turnaround time, reduction in errors, better efficiency, better customer experience, better scalability, automation, the number of users affected and so on.
3. List down the constraints - It is important to understand how ready your organization is and list the possible constraints and complexities in solving this problem. Some of them could be the relevance of the data available, the amount of data required, availability of required skill set, domain knowledge, technical knowledge etc. Accordingly, these can be solved by upskilling your employees, hiring data scientists, and collaborating with AI firms who have solved similar problems in the past.
4. Prioritize business problems - After analysing the impact and the constraints, rank the business problems. Begin by focusing on those problems that offer maximum value and involve minimal complexities and constraints.
5. Solution building – Once you start working on the problem, gather more insights into the information required to solve the problem. Also, make sure that the team understands the problem, it's business impact, and has relevant knowledge. After this, collect the relevant data and build the model. Later, test the model and perform iterations until the model gives the expected results.
6. Measure the results – This has to be done at every step to ensure that there are no or minimal deviations in the results expected at every step. Once the solution is built, measurement of the results in terms of value should also be done. Assess whether you achieved what you had planned at the end. If not, understand the causes and work on them.

AI accelerates the journey of enterprises towards becoming intelligent enterprises. The concept of intelligent enterprises focuses on making the way we work seamlessly. By incorporating AI/ML in various processes, our enterprises can learn to identify, plan and prioritize opportunities, derive insights, and provide appropriate recommendations to help reach the goal in the most efficient manner.AI is supported by continuous learning, which makes it more and more intelligent with experience just like humans. In this way, we gradually move towards a data-centric world transforming the way in which businesses function.

Reference

¹https://www.linkedin.com/pulse/how-netflix-uses-ai-data-conquer-world-mario-gavira

²https://yourstory.com/2019/06/amazon-fashion-feature-artificial-intelligence-stylesnap

³https://www.artificialintelligence-news.com/2019/07/09/idc-half-ai-projects-fail-companies/