Gartner: accelerating AI deployments — paths of least resistance

Gartner Analyst, Melissa Davis, outlines how data and analytics leaders can remove the most common barriers to production and accelerate from AI prototypes to scalable solutions.

This is a contributed article by Melissa Davis, Senior Director, Analyst at Gartner.

As the world progresses into a ‘new normal' and the disruptions to global health and economic recovery remain, the need for innovative artificial intelligence solutions has increased and faster results are now required.

Gartner predicts that by the end of 2024, 75% of organisations will shift from piloting to operationalising AI, driving a 400% increase in streaming data and analytics infrastructures.

The health and economic disruption caused by COVID-19 has increased the pressure to accelerate AI deployments from proof of concept (POCs) to scalable production implementation. The objectives are to realise the value of these deployments and deliver on their promise, while managing inflated expectations. Gartner found through a recent research study that today only 53% of POCs make it into production, taking an average of 9 months.

Gartner outlines how organisations can remove the barriers and accelerate AI deployments from POC into production.


Accelerating artificial intelligence proofs of concept into production

Moving from hype to reality, AI has been proven to deliver tangible and measurable business value. Gartner's 2020 AI in Organisations Survey found that contrary to popular belief, a lack of AI talent is not the main barrier for successful deployment of AI techniques, with the paths to scale AI solutions from POC to production falling into three main categories:

  1. Achieve the business value from AI use cases:
  • Align AI investments to strategic objectives, relentlessly measuring business value and outcomes.
  • Start small with a few use cases that matter and will show value. Drop the outliers.

  1. Establishing diverse, multidisciplinary teams:
  • Engage with both the business and IT early on during POCs and throughout the project life cycle.
  • Build multidisciplinary roles in the AI team.; include business domain experts and software engineers, and machine learning operationalisation (MLOps), and other disciplines.

  1. Scaling AI solutions from POC to production infrastructures:
  • Tackle the system integration, security and privacy issues by engaging with MLOps and DevOps, data security and other enterprise IT teams.


Practices to identify and optimise business value from use cases

The highest level of business value is achieved when AI initiatives are tightly aligned with strategic priorities. Organisations should begin with a noncore use case to demonstrate success, picking a couple of individual projects at most and dropping the long shots. Best practices include:

  • Start small. Prioritise core use cases that have real value to the company. Do not engage in a large complex project.
  • Choose the use cases based on the most promising areas for AI across your industry or business process domain.
  • Target AI projects based on your organisation's AI maturity.
  • Quantify the value by having a system in place for measuring the value generated by the model.


Establishing diverse, multidisciplinary teams

As revealed in Gartner's 2020 AI in Organisations survey, 70% of respondents state that a lack of AI talent is not a major barrier to successful AI deployments. Despite this, teams that don't establish a diverse team tend to fail.   

CIOs looking to establish AI teams need to determine which roles and skills will be required. In doing so, those responsible must be aware that AI is a much broader discipline than machine learning (ML). Organisations should identify and prioritise their main AI business use-case categories and required capabilities as early as possible.

Those organisations successful in operationalising AI techniques also educate their peers and foster trust in AI techniques across the entire organisation — from everyday users to the executives. This is a critical element of success.


Scaling AI solutions from proof of concept to production infrastructures

According to Gartner, the top two identified barriers for organisations when scaling production of AI solutions include the security and privacy concerns and the complexities and difficulties of integrating solutions with existing infrastructure. 

To scale solutions, CIOs and IT leaders must build competencies to integrate AI solutions within the organisation's infrastructure.

For successful scalability CIOs and IT leaders should operationalise AI techniques by involving middleware specialists and application integration specialists, and other IT application teams including DevOps and MLOps to tackle the integration complexities.


Putting Gartner's recommendations into practice

Considering the current global pandemic, organisations in most cases have had years of digital transformation happen in only a few months. This rapid acceleration doesn't come without challenges, as organisations face increased barriers in the successful adoption and acceleration of AI solutions.

CIOs and IT leaders responsible for operationalising AI can increase the chances of success by strictly measuring the business value and outcomes from AI initiatives across the organisations, tightly aligning AI investments with strategic priorities.

Secondly, success is achieved when CIOs and IT leaders build multidisciplinary teams, involving both business domain experts and IT early in the POC stage and continuously throughout the entire deployment process.

And lastly, they must address the complexities and difficulties when embedding and integrating AI with other applications and infrastructure, building competencies to ensure seamless integration of AI solutions with their organisation. This can be achieved by involving middleware specialists, application integration specialists, and other IT teams such as DevOps.

Melissa Davis is a Senior Director, Analyst, at Gartner. She advises clients on the use of customer analytics to improve the customer experience and deliver business value across digital commerce, marketing, sales and customer service business processes. Davis covers customer analytics strategies, trends and technologies.