AI technologies are quickly maturing as a viable means of enabling and supporting essential business functions. However, creating business value from artificial intelligence requires a thoughtful approach that balances people, processes and technology.
AI comes in many forms: machine learning, deep learning, predictive analytics, natural language processing, computer vision and automation. Companies must start with a solid foundation and realistic view to determine the competitive advantages an AI implementation can bring to their business strategy and planning.
“Artificial intelligence encompasses many things,” according to John Carey, managing director at business management consultancy AArete. “And there’s a lot of hyperbole and, in some cases, exaggeration about how intelligent it really is.”
What advantages can businesses gain from adopting AI?
Recent cutting-edge developments in generative AI, such as ChatGPT and Dall-E image generation tools, have demonstrated the significant effect of AI systems on the corporate world. A McKinsey Global Survey revealed a dramatic surge in global AI adoption — from approximately 50% over the past six years to 72% in 2024.
Some of the many benefits that businesses can gain by adopting AI include the following:
- Improved accuracy and efficiency in decision-making processes.
- Increased automation and productivity in business operations.
- Enhanced customer experience through personalized recommendations and interactions with chatbots and intelligent agents.
- Enhanced data analysis and insights to inform business strategies.
- Improved risk management and fraud detection.
- Cost savings as a result of process automation and optimization.
- Enhanced competitiveness and differentiation in the marketplace.
- Advanced innovation and the ability to create new products and services.
- Scalability and efficient management of large amounts of data.
- An opportunity to venture into new markets with unique AI options.
AI implementation prerequisites
The successful implementation of AI in business can be challenging. But a detailed understanding of the following factors and conditions before execution can considerably enhance the result:
- Labeling data. Data labeling is a crucial step in the preprocessing pipeline for machine learning and model training. It entails organizing the data in a way that gives it context and significance. Businesses should assess whether they have a data-driven culture within their operations and evaluate whether they have access to enough data to support the deployment of AI/ML efforts.
- Strong data pipeline. To ensure data is combined from all the different sources for rapid data analysis and business insights, organizations should strive to build a solid data pipeline. A strong data pipeline also offers reliable data quality.
- Data quality. Before training an AI model, organizations should evaluate and enhance their data quality as it affects the accuracy and efficacy of the trained model. Evaluating and enhancing data quality involves cleaning and preprocessing the data to remove errors and inconsistencies, and ensuring the data is unbiased and accurately reflects real-world scenarios. For example, when predicting customer churn, the data must represent a range of customer behaviors. Insufficient data might require businesses to generate synthetic data, which could lead to less accurate results.
- The right AI model. The success of any AI implementation can be seriously hampered by the choice of AI model a business uses. A large data set combined with an inadequate AI model could produce a large amount of training data that the model is incapable of processing efficiently. This can lead to issues such as overfitting or underfitting. Therefore, selecting the right AI model is imperative before implementing an AI strategy.
- Integrating AI into existing systems. Organizations often struggle to incorporate AI into their current infrastructure, especially with legacy systems. APIs can help overcome this struggle as they enable new AI tools to access existing data without overhauling the entire system. Middleware further helps with AI integration by acting as an intermediary that facilitates communication and data exchange between legacy systems and modern AI applications. Embracing digital transformation, such as upgrading legacy systems to cloud-based architectures, can also help achieve effective AI integration.
- AI implementation roadmap. Before starting an AI implementation, outline the AI implementation’s market launch and how its success will be measured. The roadmap should detail the execution steps, the support required at each stage and the KPIs to assess success.
13 steps to AI implementation
Early implementation of AI isn’t necessarily a perfect science and might need to be experimental at first — beginning with a hypothesis, followed by testing and measuring results. Early ideas will likely be flawed, so an incremental approach to deploying AI is likely to produce better results than a big-bang approach.
The following 13 steps can help organizations ensure a successful AI implementation in the enterprise.
1. Build data fluency and understanding
Practical conversations about AI require a basic understanding of how data powers the entire process. “Data fluency is a real and challenging barrier — more than tools or technology combined,” said Penny Wand, executive coach at LAH Insight LLC. “Executive understanding and support will be required to understand this maturation process and drive sustained change.”
2. Define your primary business drivers for AI
“To successfully implement AI, it’s critical to learn what others are doing inside and outside your industry to spark interest and inspire action,” Wand explained. When devising an AI implementation, identify top use cases, and assess their value and feasibility.
In addition, consider who should become champions of the project, identify external data sources, determine how you might monetize your data externally and create a backlog to ensure the project’s momentum is maintained.
3. Identify areas of opportunity
Focus on business areas with high variability and significant payoff, said Suketu Gandhi, a partner and chair of strategic operations at digital transformation consultancy Kearney. Teams comprising business stakeholders who have technology and data expertise should use metrics to measure the effect of an AI implementation on the organization and its people.
4. Evaluate your internal capabilities
Once use cases are identified and prioritized, business teams need to map out how these applications align with their company’s existing technology and human resources. Education and training can help bridge the technical skills gap internally, while corporate partners can facilitate on-the-job training.
Meanwhile, outside expertise could accelerate promising AI applications.
5. Provide employee training and support
Organizations should invest in change management strategies to address employee concerns and resistance to AI adoption. This involves engaging employees early on in the process and offering them ongoing support and training during the transition.
Providing comprehensive training on AI concepts, AI-powered tools and their specific applications will help employees understand the technology, appreciate its benefits and alleviate any apprehensions they might have. Additionally, executives and team leaders should actively participate in AI initiatives, demonstrating their commitment and encouraging employees to engage with the technology.
6. Select the vendors and partners
Vendor and partner selection for AI implementation is a crucial step for organizations. When selecting vendors, companies should explore those with relevant industry expertise and a proven track record in similar AI projects. This ensures they can deliver measurable results.
It’s also important to assess the technical capabilities of potential vendors to ensure their methods are compatible with existing systems and will scale well in the future. Vendors interested in long-term partnerships should be considered as they are most likely invested in mutual success.
Due diligence should be conducted when selecting vendor candidates by checking references and evaluating their financial stability. Once an AI vendor is selected, the company should present clear service-level agreements during the negotiation process to avoid misunderstandings and maintain accountability throughout the partnership.
7. Identify suitable candidates
It’s important to narrow a broad opportunity to a practical AI deployment — for example, invoice matching, IoT-based facial recognition, predictive maintenance on legacy systems or customer buying habits. “Be experimental,” Carey said, “and include as many people [in the process] as you can.”
8. Pilot an AI project
To turn a candidate for AI implementation into an actual project, Gandhi believes a team of AI, data and business process experts is needed to gather data, develop AI algorithms, deploy scientifically controlled releases, and measure influence and risk.
9. Establish a baseline understanding
The successes and failures of early AI projects can help increase understanding across the entire company. “Ensure you keep the humans in the loop to build trust and engage your business and process experts with your data scientists,” Wand said.
Recognize that the path to AI starts with understanding the data and good old-fashioned rearview mirror reporting to establish a baseline of understanding. Once a baseline is established, it’s easier to see how the actual AI deployment proves or disproves the initial hypothesis.
10. Measure the ROI
To evaluate the effectiveness of AI implementations, organizations must measure the AI initiative’s ROI. To achieve this, they must first set clear KPIs that align with their business objectives. Cost savings, revenue growth, customer satisfaction and operational efficiency are important metrics to monitor, as is user engagement, which can also be a sign of successful integration.
Qualitative metrics, such as enhanced product quality and innovation, should also be considered.
11. Scale incrementally
The overall process of creating momentum for an AI deployment begins with achieving small victories, Carey reasoned. Incremental wins can build confidence across the organization and inspire more stakeholders to pursue similar AI implementation experiments from a stronger, more established baseline. “Adjust algorithms and business processes for scaled release,” Gandhi suggested. “Embed [them] into normal business and technical operations.”
12. Guide overall AI capabilities to maturity
As AI projects scale, business teams need to improve the overall lifecycle of AI development, testing and deployment. To ensure sustained success, Wand offers three core practices for maturing overall project capabilities:
- Build a modern data platform that streamlines how to collect, store and structure data for reporting and analytical insights based on data source value and desired KPIs for businesses.
- Develop an organizational design that establishes business priorities and supports agile development of data governance and modern data platforms to drive business goals and decision-making.
- Create and build the overall management, ownership, processes and technology necessary to manage critical data elements focused on customers, suppliers and members.
13. Continuously improve AI models and processes
Once the overall system is in place, business teams need to identify opportunities for continuous improvement in AI models and processes. AI models can degrade over time or in response to rapid changes caused by disruptions.
Teams also need to monitor feedback and resistance to an AI deployment from employees, customers and partners.
Common AI implementation mistakes
Businesses that neglect to take the recommended steps when deploying AI risk committing the following mistakes:
- Adopting too many tools simultaneously.
- Having unclear business objectives.
- Ignoring privacy and security concerns that come with AI.
- Not collaborating with the right partners.
- Not involving stakeholders and affected employees in the decision-making process.
- Relying too much on the black box models of AI.
- Not performing enough testing and validation.
- Overlooking change management.
- Underestimating the complexity of AI.
- Neglecting ethical considerations.
What are the key challenges in implementing AI in an organization?
During each step of the AI implementation process, problems will arise. “The harder challenges are the human ones, which has always been the case with technology,” Wand said.
Penny WandExecutive coach, LAH Insight LLC
A steering committee vested in the outcome and representing the firm’s primary functional areas should be established, she added. Instituting organizational change management techniques to encourage data literacy and trust among stakeholders can go a long way toward overcoming human challenges.
“AI capability can only mature as fast as your overall data management maturity,” Wand advised, “so create and execute a roadmap to move these capabilities in parallel.”
Key challenges that organizations typically face during an AI implementation include the following:
- Data management challenges. Data management challenges include ensuring high data quality — accuracy, completeness and timeliness — to achieve effective AI performance. Poor data quality can lead to biased results, requiring strong data governance. Integrating data from various sources, especially legacy systems, can also be complex.
- Model governance. Model governance is crucial for maintaining AI reliability and ethical standards. Organizations need frameworks for security, testing and ethical compliance, and must manage version control and data lineage to ensure models are based on trustworthy data.
- Performance consistency. Maintaining consistent AI model performance is crucial, especially at scale. Variability in model performance can arise from changes in data inputs or shifts in underlying business processes. Organizations should use machine learning operations practices for repeatable model development and deployment, including regular performance evaluations and updates based on new data and business advancements.
- Integration with existing systems. Integrating AI implementation with existing systems such as CRM or ERP can be complex and often requires significant adjustments to legacy infrastructure.
- Determining intellectual property ownership. Determining ownership of AI-generated or AI-assisted outputs can be challenging, especially when multiple human and machine agents are involved. Businesses must address the risk of intellectual property rights infringement or misappropriation, including unauthorized uses of AI systems such as copying, reverse engineering and hacking.
- Effective utilization of LLMs. Finding the ideal mix between LLMs and human expertise to produce good quality, compelling and SEO-friendly content is an enormous challenge for organizations using AI. While ignoring AI technologies can reduce productivity and competitiveness, relying too much on AI can lead to poor content and plagiarism threats. To combat this challenge, businesses should thoroughly evaluate their processes to establish the optimal combination of AI and human input.
- Customer trust. Customer acceptance challenges can arise if an organization isn’t transparent with its AI implementation, which can raise concerns regarding data privacy and trust in AI-decision making process. Businesses should be transparent about their AI use, focusing on data security and demonstrating how AI complements human expertise rather than replacing it.
- Shortage of AI skills. A key challenge in AI implementation is the shortage of skilled professionals with expertise in data science, machine learning, programming and domain knowledge. To address this, businesses can invest in upskilling their current workforce through training programs and workshops.
How can businesses ensure ethical AI implementation?
Responsible use of AI technologies is becoming increasingly important as AI systems are rapidly integrated into various sectors. For instance, a healthcare organization developing an AI tool for diagnosing medical conditions could assess the tool’s potential effects on patient privacy, consent and equity beforehand. This assessment would involve reviewing how patient data is collected, stored and used, ensuring the AI tool doesn’t reinforce existing biases or produce unequal health outcomes across different patient groups.
Organizations can address ethical and governance issues surrounding AI by establishing robust governance frameworks and addressing potential risk factors such as bias, discrimination and privacy violations.
Here are several practices organizations can adopt to ensure ethical AI implementation:
- Create and execute strategies for bias mitigation, such as training AI models on diverse data sets and regularly assessing them for fairness to help with AI discrimination.
- Ensure AI systems are transparent, explainable and auditable, so stakeholders can understand the decision-making processes.
- Compliance with regulations such as GDPR and CCPA should be taken into account, as these laws not only set standards for data protection and user privacy but build trust with consumers.
- Set up clear and ethical standards and guidelines for AI development and use.
- Involve diverse stakeholders in the AI development process to address various perspectives and concerns.
- Foster a culture of organizational awareness and accountability by training employees in ethical AI practices and encouraging them to identify and report ethical risks.
- Incentivize ethical behavior within the organization to further reinforce the importance of responsible AI use.
Editor’s note: This article was updated in September 2024 to add additional AI implementation steps, provide updated survey information and improve the reader experience.
Kinza Yasar is a technical writer for WhatIs with a degree in computer networking.