- November 6, 2019
- Posted by: webo
- Category: Digital Transformation
Artificial Intelligence (AI) becomes intelligent
Artificial Intelligence (AI) continues to evolve and is becoming more intelligent. Hence it becomes critical for all of us to question its power and role in our day to day lives.
AI was initially conceptualized to help organizations streamline business processes in order to improve its efficiency and acquire new/retaining existing customers. The payoff for AI is so tempting that many organizations who haven’t yet incorporated AI are tempted to make their first move. Companies have started to believe that AI as a tool will be vital to ensure competitive advantage in order to be relevant in this ever changing world so that they are not left behind their competitors.
Inspite of AI’s potential in delivering benefits to humans, it can also hinder it. Hence companies need to demystify AI to understand and approach it with both skepticism and optimism.
The main issues companies face while moving towards integrating AI in their operations is when leaders are not clear on what it means to adopt AI. In addition, companies are not carefully balancing customer privacy and in providing customer value. In order to address the above problems, leaders need to evaluate the strategic challenges of bringing AI as an integrated element of a process and not consider AI just as a tool.
Three challenges to Artificial Intelligence (AI) implementation
Leaders do not have the clarity on adoption of AI
Many companies feel the pressure to adopt AI as it has become a buzzword and panacea for all problems. Hence most of them adopt AI without thinking through the why and how. Some leaders jump into gathering data and building models to replace or augment a business function without systematically planning for the budget and time required to invest in the solution along with a contingency plan if things do not go as planned.
In other cases, enterprises adopt new AI functions based on the data available with them. For instance a company can decide to build a chatbot based on the data samples of the recent customer service calls. This can lead to wrong conclusions as data samples needs to be clustered by time of the day, customer usage matters and by other multiple parameters using a disciplined approach in order to come up with an effective solution.
It is important that companies understand their own data as it can provide them with a competitive advantage. AI needs to be embedded with data as a part of strategic annual plan, in the context of identifying competitors, evaluating offerings and developing short term and long term strategies. AI should be a point solution to a specific identified problem rather than considering it as the end-all solution.
Garbage in and Garbage out in AI
AI has a hidden bias which can turn valuable data into decision making mistakes and false truths.This is because the complexity of AI algorithms can make it impossible to validate all possible outcomes over all possible scenarios. So when these algorithms get deployed widely both integrated within operational processes as well as to enhance customer value, there is an high risk of a negative bias in the AI decision making ability.
For instance in the Gmail smart compose AI feature which offers suggested text for finishing answers and responding to mails, there is a hidden bias of assuming that the response needs to be for a male. Though Google had rectified this issue by eliminating gender based replies, it still does not uncover the root cause of the issue that might still be present from training on data sets.
Microsoft released its famous chatbot Tay which became controversial as Twitter users engaging with Tay had taught it to develop it’s own form of racist tweets in US. This confirms the fact that if AI actively learns from garbage then it is going to spew out the same garbage as truth.
Hence it is difficult to train new algorithms with the data they collect. This is a challenge that leaders need to pay attention to.
Balancing customer privacy and customer value
Data privacy defines the relationships between collection and dissemination of data and expectation of customer privacy. Companies collect data about the customers so that they can target right ads/ products / services to the customers to improve their customer experience and also be more efficient.
At the same time there are ethical concerns regarding collection of data especially related to private data like health etc. For instance, a company can collect health data to identify patterns which can help in right diagnosis. At the same time companies can scrub the same health data to target ads or sell to other firms to generate a financial profit.
Companies need to balance the customer privacy with the value they are providing to customers with the AI offering. If this balance is not achieved, then this may result in harsh regulations that would limit the freedom to collect the relevant data.
Companies need to validate their data and be open to trying several different Artificial Intelligence (AI) methods and assessment methods. Teams need to ask the right questions when building the Artificial Intelligence (AI) models so that they can come up with AI solutions that addresses customers painpoints.
When building Artificial Intelligence (AI) solutions, different stakeholders and perspectives needs to be involved to develop clear ethical standards for the use of AI and its value proposition.