The T Minus 10 Of AI - Getting Started with the AI Journey
Kellton Tech is a CMMI Level 5 and ISO 9001:2015 certified global company, providing cutting - edge digital transformation solutions & services in strategy, consulting, digital, and technology. The organization has global workforce of 1400 employees that work together with clients - startups, SMBs, and Fortune 500 businesses.
The future belongs to Artificial Intelligence. Today people are using AI in one or the other form in their daily lives. But, when it comes to using AI in the business processes, most of the CIOs struggle to start. The reason for the struggle is different for every individual. For example, the business development department when not aligned with the IT department face difficulties in identifying visible use cases where AI can be used or calculating the ROI for the AI implementation. They also find it difficult to convince the management and allocate budgets for the AI projects. Whatever may be the challenge, ultimately organizations are losing their precious time, which they can utilize for future proofing the organization’s growth.
To get started with the AI journey, organizations must keep the following things in mind.
1. AI is not a Magic- Be practical
Don’t think that AI is the answer to everything; instead, think about its practical usage. In my experience, many CIOs whom I have met ask how they can replace a human with AI bots. Such thoughts are amateurish because you cannot replace human everywhere using AI. You can only use AI to assist humans and enable them to get more productive and accurate. Other than that you can define a standard operating procedure for AI-assisted workflows.
"Artificial intelligence offers opportunities for businesses to gain insights, maximize productivity, and ensure informed decision-making"
2. Get Business Aligned with IT
The biggest challenge I see is disconnect between IT and the business departments. If you want to succeed in your AI journey as an organization then the IT and the business department must come together to meet a common goal. I have seen business and IT departments running similar AI initiatives but working in silos which most of the times lead to failed AI initiatives.
3. Find out a Pilot Use Case for AI
Finding a pilot use case for AI is the most difficult thing. Most of the times, I have seen CIOs choosing the easiest use case for pilot irrespective of the impact that the use case might have on the organization. In my view, CIOs must choose the highest impact but low hanging use case. Choosing the highest impact use case will motivate the team and will also show case the ROI.
The best way to choose a use case is to find out things that you might want to know/achieve for your business or your customers for better planning of your business goals, providing better customer experience etc. You also need domain expertise of your business people for choosing a use case.
4. Implement the Pilot Use Case and Track it Closely
Implementation and monitoring is the most crucial phase for any AI initiative/pilot. It requires a lot of patience to get the desired results, as the implementation of AI is an iterative process.
There are many hurdles in the implementation, for example, training data quality, lack of metadata, and lack of domain and technical expertise. Organizations usually do not have the right data to start the AI pilot for the selected problem. Whatever is the case, don’t give up. Try to overcome the hurdles or make an alternate arrangement in your standard operating procedures.
A few months back, we were at the same crossroad facing similar hurdles in one of our implementation. The data was not digitalized; it was mostly on paper. We tried some OCR software’s to digitalize the data but the accuracy was less than 35 percent. Finally, we decided to manually input the data in the CSV format with the help of data entry operators. After we had the data, it was cleaned. It took us a month to achieve our goal, but we were able to do it at a much lesser cost than the technical solution.
Once the cleaned data is available, organizations can start AI model building and training and iteratively test it until they get the desired results. Be reasonable in the expectations that you might have from the results.
5. Assess the Impact and ROI
Assess your ROI in the long term and not in the short term. It is difficult to achieve 100 percent success often but if you are able to achieve 70 percent then it is worth it. Define an SOP for the cases that failed and keep the corrected data for training and tweaking the AI model. Decide the KPIs for the assessment of ROI and track them closely to achieve what you desire
6. Expand your Scope and Generalize the Idea
Take lessons from the pilot and start other pilots or generalized implementation according to your assessment.
Artificial intelligence offers opportunities for businesses to gain insights, maximize productivity, and ensure informed decision-making. In order to become an AI superhero, businesses need to follow the six steps and achieve the desired results.