Kyndryl’s Carolyn Prior discusses the importance of ethical data management before launching AI at scale, along with the challenges workers in this sector face.
The AI sector is ramping up, with industries becoming more eager to adopt its capabilities into their services.
As examples like ChatGPT have shown, many tech companies will quickly jump on potential AI opportunities, to provide better services and get an edge over the competition.
However, the implications that AI poses for our society means the “onus is on companies” to embrace this technology fast and “bring better experiences for their customers”.
That’s according to Carolyn Prior, the data and AI practice leader for Kyndryl in the UK and Ireland. In an interview with SiliconRepublic.com, Prior said this pressure to deliver new AI products at such speed “doesn’t allow space for trial and error”.
This can cause a struggle in AI development, as Prior said these systems need to be “programmed to be ethical, transparent and secure from the offset”.
“AI needs to be fed with data, and it’s critical that it is fed clean, quality data,” Prior said. “The provision of high quality data is an ongoing challenge for many organisations, with complexity at both a technological and organisational level presenting obstacles to progress.
“Data governance, management and operations become increasingly important to pave the way for successful and ethical AI initiatives.”
Prior said organisations need to first prioritise “the evaluation and organisation of their data” before they begin implementing AI at scale.
“It’s critical to undertake a thorough discovery phase before embarking on any kind of data transformation initiative,” Prior said.
In 2021, AI ethics expert Joanna J Bryson spoke to about the challenges of regulating AI and why more work needs to be done.
New challenges for AI workers
While AI was once a specialist area, Prior said these types of roles will become “increasingly mainstream” within organisations.
This combines with the “continuous challenge” workers in AI and machine learning roles face, as Prior said they have to constantly upskill themselves to “keep pace with the rate of change, with emerging technology and relevant new use cases”.
As AI becomes more mainstream, Prior believes that scaling it across an enterprise, reducing time to insight and reducing time to value will become “key success criteria for businesses”.
Workers within these positions will play a “critical role” in delivering these results for a business. As such, Prior said they will need to “stay close to the business and really consider and focus on how AI can contribute to business outcomes”.
“Today, AI projects can fail because they are not built to scale or integrated with business workflows,” Prior said. “Scaling AI and successfully integrating it across the enterprise will continue to be one of the biggest challenges facing workers stepping into AI and machine learning roles.
“They will need the support of modern data management and governance practices in their organisations to assist them in this challenge.”
Prior said its also important for organisations to create cultures where workers can raise concerns about AI systems “without stifling debate and innovation”.
“A foundation of organisational trust in data quality and stewardship, built on strong governance practices, is an important start,” she said.
Exciting trends ahead
Prior said AI-driven automation is a central part of Kyndryl’s “advanced delivery strategy”, which is helping the company become a managed services partner for customers that are moving to the cloud.
As traditional IT services become more automated, Prior said this is having a positive impact on Kyndryl’s incident management, application deployment, security and compliance services.
“It’s a proactive way to operate which reduces risk and results in fewer incidents and faster resolution,” Prior said.
Prior said Kyndryl recently worked with Irish energy supplier Bord Gáis, by offering automated services that can “proactively resolve repetitive tasks and automate event incident response”.
She said the energy supplier’s IT security patching cycle previously involved labour-intensive six-week cycles every quarter, but automation cut this work down to two days every month.
“As we mark 15 months as an independent company, I’m really excited about what comes next for Kyndryl and our customers,” Prior said.
10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.