There has been an unavoidable buzz around artificial intelligence in recent years, with commentators quick to highlight both the positive and negative aspects of AI technology. Businesses have been quick to try to adopt AI and attempt to reap the promised benefits of innovation, efficiency, and productivity , with research from IBM’s 2024 ‘Global AI Adoption Index’ showing that over half of companies surveyed (59%) are exploring or deploying AI and plan to accelerate investments in this area. But the rush to gain a competitive edge has also been held back in some industries by one major concern, whether AI adoption will wipe out certain human roles?
This is a misconception – by understanding that AI is not an entity, but a tool that can only work as well as the humans that enable it, organizations can deploy AI more effectively and with much more control. We only need to look at the recent high profile AI-based order systems rollout failure of McDonalds. The fast food chain announced it will discontinue its AI drive through solution in more than 100 restaurants due to multiple mistakes in food orders, which then required human intervention for approximately one in five orders to rectify issues. As this case demonstrates, AI has its limitations and human skills will continue to be critical - anyone looking to deploy AI to replace employees will face numerous challenges. Instead, they should embrace the technology as a support system for menial tasks, streamlining repetitive processes and taking some pressure off end-users.
Here are a few areas of focus for organizations considering AI deployment:
AI doesn’t need to be revolutionary; it needs to be practical
Oftentimes, organizations jump the gun on deploying new tech as part of a transformation path, only to find that they are overspending on a technology that, six months down the line, is failing to generate a return on investment. This is typically because they are too ambitious or underestimate the journey ahead of them.
Rather than throwing millions into the loose promise of improved productivity, organizations need to first sit down, establish realistic goals and evaluate where AI can support their people and how it can be incorporated into their business objectives.
Doing a thorough analysis of processes before embarking on any AI project is key to evaluating which part of the business AI will function best and deliver the strongest value. Whether the objective is to improve internal processes or deploy to external users, questions such as ‘who will be using this capability?’, ‘to what scale does it need to function?’, and ‘how are we measuring success?’ form fundamental expectations and help to keep projects focused.
Sometimes this means automating a singular process, such as customer service inquiries. This small-step approach is vital to ensuring and measuring success. But AI won’t be beneficial to all departments, nor all operations – particularly if a process runs on too small a scale to justify investment. Interrogating need and value from the outset with an initial assessment is the most efficient approach to any transformation, ensuring that AI investments are meaningful and almost guaranteed to be worthwhile.
Similarly, defining success requires forward planning. Once the most effective AI use case has been established, it’s critical to determine baseline metrics for productivity and quality. This offers a clearer picture of the true impact of a deployment on operations, allowing AI investments to be assessed, validated, and learnings recorded for future projects. AI is not going away, but many people are right to be cautious. If an organization wants to adopt the technology widely, fostering a culture through demonstrating success on a smaller scale will go a long way towards a smooth transformation. If AI is to truly make life easier for its end-users, focusing on practicality and having a firm plan in place is pivotal to success.
Understand and foster AI maturity
Once the groundwork has been laid for AI success, it is crucial to map objectives against an accurate picture of an organization's AI readiness. As a new and powerful technology, AI requires skills and an ability to manage and protect complex infrastructure. Take stock of the organization's technological foundation by assessing the existing technology stack. Can current software handle the demands of AI integration? Are there any compatibility issues or technical weaknesses that will need to be addressed throughout this project?
But this isn’t just a technical question: there is a critical human element to AI maturity. Create a matrix of the skills and knowledge of both IT professionals and the team who will be working directly with the new AI program. It is common for there to be gaps here, AI is, of course, a strikingly new technology for enterprises. But if users do not know how AI works and how to use it within their roles, any investment will be practically useless. Invest in building technical confidence up front, making sure they fill any gaps with adequate training or recruitment.
The final factor to consider here is process discipline. Ensure that incoming AI deployments will be met with well-defined processes to handle data collection, management, and model development. Without all of these elements in order, organizations may be looking at a recipe for disaster.
Keeping security and quality at the forefront
Any new technology introduces a level of security risk to a business, sometimes by complicating data protection, introducing code vulnerabilities, or widening the existing attack surface. Some weaknesses may only surface once the technology is in use, but in the development stage, security should be continually built into the program and assessed with regular code reviews. As soon as discussions start about launching a new product or program, organizations must also discuss their strategy to build and deploy it securely. This is especially important in the case of AI, which brings complexity and processes masses of potentially sensitive data. Organizations developing AI models for their business or buying AI solutions from partners must stay diligent about the quality of the training data.
Beyond data quality controls, organisations must develop customised AI use policies, making sure they not only comply with government or industry regulations but also address the specific needs of the business. Having a comprehensive set of guidelines and safeguards in place for end-users significantly reduces AI misuse. These policies, coupled with proper training and process discipline will ensure that organizations can benefit from AI’s many positives whilst mitigating cyber risk.
As AI sweeps through the technology industry, it can feel like the start of a new era. But organizations must remember the basics of deploying any new technology – AI is not so different. By taking the time to truly analyze needs and plan projects to bring real benefit to the business, any organization can set themselves up for success from the get-go.
We've featured the best AI phone.
This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
Most Popular
Comments / 0