Should You Be Building or Buying AI/ML Solutions?
Most businesses are building AI solutions from scratch. But is this the best way forward?
Artificial intelligence and machine learning (AI/ML) solutions are creating infinite opportunities for organizations to evolve. AI/ML is a rapidly growing technology area with generally positive perceptions as we discovered in our 2022 Rackspace Technology AI/ML Annual Research Report. The report surveyed 1,870 IT leaders and found that 70% of respondents believe AI/ML is having a positive impact on brand awareness, company reputation, revenue generation and expense reduction.
The current state of AI/ML
AI/ML is considered to be a critical element of business strategy. Our AI infographic, based on our report, reveals that AI/ML is now perceived as the most important technology for strategy by 64% of respondents, placing it above cybersecurity and cloud computing.
These positive sentiments explain why we are seeing organizations increase budgets for AI/ML initiatives. In 2021, there was a 55% increase in companies investing over 6% of their entire IT budgets in AI/ML initiatives compared to 2020, according to our report.
I believe this is a consequence of a growing familiarity with AI/ML as information relating to it becomes more readily available. This information is helping to fuel organization’s confidence in AI/ML and encourage investment. In addition to this, organizations are now processing data in ways that makes it easier to use in AI/ML initiatives.
But it's not all positive news, as Gartner predicts that 85% of AI projects will deliver erroneous outcomes due to bias in data and algorithms, which are influenced by the managing teams. This would imply that organizations require experienced internal managers or partners to guide them in the building of AI/ML solutions.
Organizations are choosing to build AI/ML solutions
Despite the potential for errors, according to our report, organizations are keen to build AI/ML solutions. Sixty six percent of survey respondents said AI/ML technology is past the experimentation stage in their organization and is now either in “optimizing/innovating” or “formalizing” states of implementation.
Our report also revealed that 53% of IT leaders said they'd rather build an AI/ ML solution from scratch. I believe this enthusiasm to take a DIY route stems from a preconception that AI/ML solutions take longer to implement when using components that are bought. Developers have limited knowledge of the more complex components, so it takes longer to implement them. It is also a challenge to keep up-to-date with the latest offerings for AI/ML.
Why it makes sense to buy AI/ML solutions
For simpler AI/ML initiatives, it’s already easy to implement components from one of the hyperscale cloud providers, like Google Cloud Platform™ or AWS. Over the last five years, hyperscale cloud providers have made it a lot easier for organizations to use AI/ML related services and tools. Previously, you would’ve needed a PhD to be able to use AI/ML products but now there is click-and-go functionality.
I think there will be a shift toward buying the components of an AI/ML solution with organizations using external platforms or off-the-shelf services or tools more frequently. As organizations become used to deploying AI/ML solutions the confidence to buy components will increase. We saw the beginning of that boom in 2020.
By buying off-the-shelf services or tools, which already incorporate automation, organizations can reduce costs and save time. I’ve noticed that there are often certain stages in an AI/ML project that require the same actions. It is these stages that are automated in partner tools and services. By using pre-built components, developers can spend more time working on customizations. For example, AWS has some great services that can take away up to 90% of the team’s work, freeing them to focus on the more creative aspects of the initiative.
The benefits of using cloud
In my opinion, using a cloud platform for an AI/ML project enhances cyberdefenses because security updates are automated. This level of protection is difficult to achieve in a built from scratch AI/ML solution. When an organization builds an AI/ML solution by itself security protocols must be implemented step-by-step and it all has to be connected. In this scenario, it is easy to miss things.
The case for utilizing a cloud platform is proven by McKinsey research that suggests AI high performers could be gaining some of their efficiency from cloud platforms. In the McKinsey survey, the organizations identified as AI high performers run 64% of their workloads on public or hybrid cloud and access a wide range of AI capabilities such as natural-language-speech understanding and facial recognition.
The move toward buying, rather than building, components of AI/ML solutions will also be driven by a shortage of talent. At present, 86% of IT leaders are actively seeking to hire AI/ML experts, according to our report. The most challenging area to hire for is ML, followed by collaborative robotics, deep learning and predictive modeling. The success of AI/ML programs is being hindered by a lack of skills with more than one-quarter of organizations reporting that they are unable to fully exploit the results of AI/ML initiatives.
Experienced partners enable success
Many organizations are, to some degree, working with an experienced provider to navigate the complexities of AI/ML development. Partnering with two or more companies for AI/ML initiatives is a feature of 83% of high-achieving organizations, according to Deloitte. Taking this route gives organizations access to expertise and technology that can accelerate development and increase the overall success of a project.
It seems that although many organizations are keen to build their own AI/ML solutions there will be a trend to buying AI/ML components off-the-shelf. I feel as organizations gain confidence in AI/ML development and deployment they will become more comfortable with buying services and tools to incorporate in solutions. The future for AI/ML services, tools and solutions is a bright one.
AI/ML Annual Research Report 2022
About the Authors
Professional Services Delivery Architect
Miriya Molina
Miriya Molina has 10 years of experience designing, developing and leading technology teams delivering the future in Artificial Intelligence, Machine Learning and Data Science. She has a creative passion for solving interesting problems and has contributed to a diverse range of industries including finance, real estate, and energy technology. Prior to Rackspace Technology, Miriya was Vice President of Product and Chief Technology Officer across different startups where she implemented pattern recognition systems and delivered end-to-end technology solutions. She is currently a rising star at Rackspace Technology within the Onica Group.
Read more about Miriya Molina
