Intro
A survey by Deloitte found that 68% of executives believe AI will be a strategic priority for their businesses in 2020. (source: https://www2.deloitte.com/us/en/insights/focus/cognitive-technologies/state-of-ai-and-intelligent-automation-in-business-survey.html ). That was almost three years ago, before all the news of ChatGPT, and other AI tools that had all the media abuzz. This means that Artificial Intelligence (AI) is no longer a buzzword but a transformative technology that has the potential to revolutionize the way organizations work. While it has a great ability to analyze data and provide insights, AI is different from analytics. Analytics help understand data, AI helps predict the next step., AI can help businesses make informed decisions, automate mundane tasks, and improve productivity. However, most IT departments underestimate what it takes to put it into production and its implications, leading to a misguided implementation that fails to deliver expected results. In this article, we will explore why most IT departments are misunderstanding AI and how they can properly take advantage of it.
The Hype around AI
The hype around AI has led to a rush of companies eager to jump on the bandwagon. According to a survey by Gartner, 54% of CIOs globally say that AI initiatives are in production (source: https://www.gartner.com/en/newsroom/press-releases/2021-06-15-gartner-survey-finds-75–of-organizations-invest-in-ai) or are in the process of being deployed, up from 35% in 2018. However, despite the enthusiasm surrounding AI, most companies fail to see the bigger picture and end up investing in AI without proper planning.
One of the main reasons for this hype is the notion that AI is a magic bullet that can solve all problems. Most companies believe that implementing AI will lead to better decision-making, increased revenue, and reduced costs. However, AI is not a silver bullet, and its implementation requires careful planning and execution.
Misconceptions about AI
Is it a catch all solution?
One of the biggest misconceptions about AI is that it is a one-size-fits-all solution. Most companies fail to understand that AI is not a standalone technology but a tool that can be used in various ways. Yet AI requires an understanding of the problem it is intended to solve and the data it will use to solve it.
What type of intelligence is it really?
Another misconception is that AI can replace human intelligence. While AI can automate mundane tasks, yet it cannot replicate human creativity, intuition, and empathy. AI works best when it complements human intelligence, not replaces it.
Do we really know what it is?
Another reason for the misunderstanding of AI is the lack of expertise within IT departments. AI is a complex technology that requires a deep understanding of data analytics, machine learning, and natural language processing. Most IT departments lack the necessary skills to implement and manage AI effectively.
The lack of expertise is compounded by the shortage of AI talent in the job market. According to a report by LinkedIn, AI specialist roles have grown 74% annually over the past four years, and there are only a limited number of people who have the necessary skills to fill these roles. This shortage of talent means that most companies struggle to find the right people to lead their AI initiatives.
How to Properly Implement AI
To properly implement AI, companies need to take a holistic approach that involves proper planning, investment, and execution. Here are some steps companies can take to properly implement AI:
Identify the Problem
The first step in implementing AI is identifying the problem it is intended to solve. Companies need to understand the business problem and the data they have at their disposal to solve it. Once the problem has been identified, companies need to determine whether AI is the best solution for the problem or whether other technologies can be used.
Develop a Roadmap
Once the problem has been identified, companies need to develop a roadmap for implementing AI. The roadmap should include a clear definition of the problem, the data that will be used, the AI algorithms that will be applied, and the expected outcomes. The roadmap should also include timelines, milestones, and budget estimates.
Invest in Talent
To successfully implement AI, companies need to invest in talent. This includes hiring data scientists, machine learning engineers, and AI specialists. Companies can also upskill their existing employees to fill these roles. Investing in talent is crucial to ensuring the success of AI initiatives.
Test and Learn
AI is a complex technology that requires testing and learning. Companies need to test their AI models and algorithms to ensure that they are effective in solving the problem they were designed for. Testing should involve different scenarios and datasets to ensure that the AI model performs consistently. Learning involves analyzing the results of the tests and making adjustments to the AI model to improve its performance.
Monitor and Evaluate
AI models are not static, and their performance can degrade over time. Therefore, companies need to continuously monitor and evaluate the performance of their AI models to ensure that they are still effective in solving the problem. This involves setting up metrics to measure the success of the AI model and comparing them to the expected outcomes.
Areas where AI is Misused
There are several areas where design failure can lead to poor outcomes. Here are some examples:
Chatbots
Chatbots are a popular application of AI in customer service. However, poorly designed chatbots can lead to frustrated customers and decreased satisfaction. For example, if a chatbot cannot understand a customer’s request or provides irrelevant responses, the customer may become frustrated and dissatisfied.
Predictive Analytics
Predictive analytics is another area where AI is misused. Predictive analytics uses historical data to predict future outcomes. However, if the data used to train the AI model is biased or incomplete, the predictions made by the model may be inaccurate. A very good example of this, is the historic real estate data impacted by “redlining” (https://www.fedbar.org/blog/can-a-machine-really-discriminate/). This can lead to poor decision-making and financial losses.
Hiring and Recruitment
AI is increasingly being used in hiring and recruitment to screen job applicants. However, if the AI model used to screen applicants is biased or incomplete, it may discriminate against certain groups of people. For example, an AI model may discriminate against women or people of color if the training data used to develop the model is biased.
Areas where AI can actually help
AI has the potential to benefit several areas of business when implemented properly. Here are some examples:
Supply Chain Management
AI can be used to optimize supply chain management by predicting demand, reducing waste, and improving efficiency. For example, AI can be used to predict when a product will run out of stock and alert the supplier to restock it. This can help prevent stockouts and lost sales.
Fraud Detection
AI can be used to detect fraud by analyzing patterns and anomalies in data. For example, AI can be used to detect credit card fraud by analyzing transactions and identifying unusual patterns. This can help prevent financial losses and protect customers.
Customer Service
AI can be used to improve customer service by providing personalized recommendations and resolving issues quickly. For example, AI can be used to analyze customer data and provide personalized recommendations for products or services. This can help increase customer satisfaction and loyalty.
Conclusion
AI has the potential to transform the way organizations work, but its implementation in addition to careful planning and execution requires a deep understanding of what AI can and can’t do, as well as understanding that unlike Science Fiction, AI doesn’t just get a sentence of instructions and run with it. Most IT departments misunderstand AI and its implications and impact on all the associated systems, leading to misguided implementation that not only fails to deliver expected results but could also impact existing systems and data. To properly implement AI, companies need to take a holistic approach that involves proper planning, investment, and execution. Companies also need to understand the problem AI is intended to solve, invest in talent, test and learn, and monitor and evaluate. By properly implementing AI, companies can benefit from improved efficiency, better decision-making, and increased revenue.
