As the benefits of artificial intelligence (AI) become increasingly clear, more and more businesses are eager to implement this technology into their operations. However, for large companies with complex systems and processes, AI adoption is often not a straightforward process. There are numerous roadblocks that can impede progress, including technical challenges, organizational resistance, data quality issues, and regulatory concerns. In this blog post, we’ll explore these common roadblocks in more detail and discuss strategies for overcoming them, so that large companies can fully realize the potential of AI.

Technical challenges

Poor data quality

AI models require large amounts of high-quality data to be trained effectively. However, many companies struggle with data quality issues such as incomplete, inaccurate, or outdated data, which can make it challenging to develop accurate and reliable AI models.

Fixing data quality issues can be a complex process, but companies can take several steps:

  1. Prioritize data cleaning efforts: Don’t try to clean everything at once. Start by prioritizing your AI use cases, then clean the data that the top use case needs. This will help guide any future investment in data cleaning and data management. Don’t lose the opportunity to iterate.

  2. Clean manually, then automate: Avoid the temptation to start by applying general rules, then tweaking from there. Like the point above, this is about learning as early as possible. Clean by hand to learn what problems exist in the dataset, then set up cleaning pipelines for specific problems you identify. Write down everything you find so you can address the root causes later.

  3. Plan on cleaning data for the long haul: An important point to pair with the one above: don’t clean one-off datasets. Commit to cleaning the dataset for the rest of its useful life. Avoid situations where a cleaning method can’t be reproduced, as this can lead to dependencies on a dataset with inconsistent cleanliness over time.

  4. Make clean data easy to access, find, and understand: Make it clear if data has been cleaned and how it’s been processed. Provide users with important information, like cleanliness, methodology, and production-readiness. Also estimate your data’s lifespan and document it (e.g. expiry or half life).

  5. Use data management tools: Once you understand your data cleaning and management needs, use tools to help automate processes and organize data. Some tools can also help with data quality testing. The right tool varies by company and need, so consider a range of options when choosing one.

  6. Make people accountable for data quality: Designate product owners for data, and treat production-grade data like any other product. Ownership leads to accountability, which leads to results.

Data cleaning can be a tedious process but can make or break your company’s ability to leverage AI. Start small, focus on learning, and build clear, scalable, easy-to-use systems.

Difficulty integrating AI into existing systems

Integrating AI into existing systems can be complex and time-consuming, especially if the company’s IT infrastructure is outdated or not designed to support AI applications. However, the process below may help:

  1. Prioritize AI use cases: Start small with one or two use cases that are most likely to demonstrate ROI in the near term and act as valuable springboards for larger future projects. Expect to spend plenty of time learning on the first few projects, so make sure they are minimum viable projects with strong business motivation to keep the projects going.

  2. Assess the impact on existing systems: Conduct an initial assessment of which systems will be impacted by your MVP and what the impacts will be (this includes IT and business systems). Document any potential impacts on and constraints of the current systems. Some of impacts may be specific to the AI technologies you use, so expect to revisit this as the project continues to progress.

  3. Evaluate AI solutions: Evaluate different AI platforms and technologies to identify the best fit for the company’s needs. Consider factors such as problem-fit, ease of integration, scalability, compatibility with existing systems, and the appropriate level of specificity of the technology. Also consider the level of integration support you could receive from vendor-based solutions or specialized consultants.

  4. Develop an agile integration approach: Develop an agile approach to AI integration that prioritizes learning and incremental progress. This will help the organization manage the integration process more effectively, support future integration efforts, and minimize disruption to existing systems.

  5. Repeat: Take the learnings from the last iteration and use them to guide future projects and investments in AI technologies, IT infrastructure, and business systems.

It’s important to see AI integration as a journey taken as a series of steps. By adopting an agile approach to existing system integration, companies can minimize the risk and frustration that comes with adapting existing systems to new technology.

Organizational challenges

Resistance to change

Employees may be hesitant or even afraid of AI, fearing that it may lead to job losses or a loss of control. This resistance can slow down the adoption process and make it difficult to gain buy-in from key stakeholders.

Here are some steps that may help reduce cultural resistance to adopting AI technologies:

  1. Start with awareness: Educate employees about planned AI projects and their potential benefits. This can help reduce concerns and uncertainty about the new technology and help employees understand how AI initiatives can improve their work.

  2. Involve employees: Involve employees in the AI adoption process by soliciting their feedback and ideas. This can particularly help in finding ways to help employees be more comfortable with new AI-related technologies. Employees who are more comfortable with new technology can help those who are less so, and employees may have additional ideas about how AI can help them in their jobs.

  3. Provide training: Provide training to employees on how to use AI technologies effectively. This can help reduce anxiety about the technology and help employees feel more comfortable using it.

  4. Cultivate psychological safety: Communicate clearly about the impact of AI on jobs and roles within the organization. Be transparent about any changes that may occur as a result of AI adoption and provide support to employees who may be impacted.

  5. Foster a culture of innovation: Foster a culture of innovation that encourages experimentation and risk-taking. This can help employees feel more comfortable with the idea of trying new technologies and can help create a more receptive environment for AI adoption.

Cultural change is often slow, so consider developing a long-term plan to ensure these activities continue to reduce organizational hesitation.

Lack of skilled personnel

AI requires a high level of technical expertise, and companies may struggle to find employees with the necessary skills to develop and implement AI systems. There are several approaches that companies can take to address this, and they may find a combination of these to be most effective:

  1. Develop internal talent: Identify employees within the organization who have an aptitude for AI and provide them with training and development opportunities to build their skills. This can help build a pipeline of internal talent for future AI initiatives.

  2. Partner with universities and training programs: Partner with universities and training programs to recruit AI talent and provide employees with opportunities to build their skills. This can help create a more robust talent pipeline and provide employees with access to the latest AI technologies and techniques.

  3. Outsource AI work: Consider outsourcing AI work to third-party providers who have the necessary skills and expertise. This can help the company access the skills it needs without having to build a large internal team. In addition, third-party providers may be able to bring project management expertise that can help projects deliver value more quickly.

  4. Build an internal AI community: Build a community of practice within the organization where employees can share knowledge and collaborate on AI projects. This can help build a more collaborative culture and enable employees to learn from each other.

  5. Focus on retention: Develop programs to retain AI talent within the organization, such as career development opportunities and a supportive work environment. This can help the company retain its AI talent and build a more sustainable talent pipeline.

Regulatory and ethical concerns

AI technologies can raise regulatory and ethical concerns related to privacy, bias, and accountability. Companies may need to invest in additional resources to ensure that their AI systems comply with regulations and ethical standards.

  1. Integrate data privacy and ethics considerations from the very beginning: Ensure that MVPs and pilot projects carefully consider data privacy and ethics from the start. This can help ensure that privacy and ethics are “baked in” rather than “iced on,” avoiding significant issues down the road.

  2. Develop an AI governance framework: Establish clear processes for raising and handling AI ethics and regulatory risk. Ensure accountability by clearly identifying the person or persons responsible for determining acceptable uses of AI and empower them to restrict the use of AI in situations in which they deem the ethical or regulatory risks too high.

  3. Develop an ethics policy: Develop an ethics policy for AI that outlines the company’s principles related to the use of AI technologies. This can help guide day-to-day decision-making and communicate implications of the AI governance framework to the broader business.

  4. Foster transparency: Foster transparency around the use of AI technologies by providing clear explanations of how the technology works and how it is being used. This can help build trust with stakeholders and alleviate concerns around the potential impact of AI on society.

  5. Invest in diversity and inclusion: Invest in diversity and inclusion initiatives to ensure that AI technologies are developed and deployed in a manner that is fair and unbiased. This can help mitigate the potential for AI to perpetuate existing biases and inequities in society.

  6. Engage with regulators: Engage with regulators and policymakers to understand the regulatory landscape and ensure that the company is complying with relevant regulations and standards. This can help the company avoid potential legal or reputational risks associated with non-compliance.

  7. Conduct ethical impact assessments: Conduct ethical impact assessments of AI projects to identify potential ethical and social risks associated with the use of AI. This can help the company mitigate these risks and develop strategies to address them.

By taking these steps, companies may be able to mitigate ethical and regulatory risks surrounding the use of AI and ensure that AI is used in a responsible and ethical manner.

Cost

Depending on the use case, implementing AI can require significant upfront investment; many companies will be reluctant to make this investment without a clear understanding of the potential return. These steps can help reduce the investment risk:

  1. Develop a business case: Develop a business case for AI implementation that outlines the potential benefits and ROI associated with the technology. This can help justify the cost of AI implementation and provide a clear rationale for the investment. Planning to introduce AI into a company without a clear and valuable use case is unlikely to result in successful AI adoption.

  2. Start small: Start with a small-scale AI implementation project to test the technology and demonstrate its value. This can help the company gain confidence in the technology and build expertise and momentum for larger-scale AI initiatives.

  3. Prioritize use cases: Prioritize use cases for AI implementation based on their potential impact on the business and probability of success. This can help the company focus its investment on the most promising areas and ensure that it is getting the most value from its AI investment.

By starting with clear, valuable, and attainable use cases, companies can get more comfortable with the cost of AI implementation and make informed decisions about whether to proceed with AI initiatives.

Is AI adoption worth it?

Despite the obstacles that companies may face when adopting AI technologies, there are compelling reasons why they might consider doing so.

By automating repetitive tasks and processes, AI frees up employees to focus on more complex and higher-value tasks, leading to cost savings and increased efficiency. AI’s ability to analyze vast amounts of data and identify patterns and insights that are difficult for humans to discern can also result in better-informed decisions and improved accuracy.

Companies can also use AI technologies to enhance customer experiences by providing personalized recommendations and interactions, leading to increased customer satisfaction and loyalty. In research-intensive industries, AI technologies can provide a competitive advantage by enabling companies to innovate and develop new products faster.

Despite the challenges associated with AI adoption, the benefits of adopting AI technologies are significant and can have a transformative impact on businesses. Companies that are willing to invest and experiment stand to gain a significant competitive advantage in the marketplace.