Loading...
You are here:  Home  >  Data Education  >  BI / Data Science News, Articles, & Education  >  BI / Data Science Blogs  >  Current Article

Scoop Up AI’s Benefits, Skip its Risks and Pitfalls

By   /  November 5, 2018  /  No Comments

Click to learn more about author Oksana Sokolovsky and Rohit Mahajan.

It seems everyone is awash in enthusiasm for Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL). Indeed, there’s a lot of business advantage, both tactical and strategic, to be had from the technology. But there’s potential business risk as well, and a lot of factors that can challenge successful adoption. Are these insurmountable?  No. But accounting for these factors ahead of time can save a lot of grief. In this article, I’ll point out some possible AI pitfalls to help you prepare for them, and I’ll provide advice on how to avoid them altogether.

The biggest challenge with AI is that it’s hard for a single organization to implement on its own, and individuals possessing the rarefied skill set required to do it are in short supply. Data Scientists, as these specialists are often known, need a cocktail of statistical, database, cognitive science and business domain expertise. These skills are each specialized enough on their own; requiring the combination of them constitutes a very tall order.

The Data Scientists
If you are able to compete and win in the Data Scientist labor pool contest, you’ll need to contend with the costs incurred from that victory. Ever since the Data Scientist role was deemed the “sexiest job of the 21st century” by Harvard Business Review six years ago, people in the role have commanded premium compensation. That’s a cost you’ll need to take on and one that will only grow. In addition, if your AI initiatives gain traction and the Data Science work becomes mission critical, you’ll need to grow the team and the cost will increase many fold.

At this early point in the discipline, Data Scientists are talented, curious people, with a knack for experimenting with the newest technologies. As nice as it is to have such talent on your team, it doesn’t always make for productivity. Playing with new technologies takes time and not a small amount of trial-and-error. Honestly, it may not be in your company’s interest to fund such work, but it also may be hard to avoid.

Tools Need Sharpening
On the other hand, a lack of productivity may not be an issue. Even in this optimal case, the tools Data Scientists use won’t necessarily aid their productivity. Right now, the state of the art in AI tooling involves writing code in Web-based notebooks that can be shared with colleagues. Such work is not terribly structured; it’s more academic in nature than it is industrial.

Experimentation – the process of working with different algorithms and parameter values, then measuring results, and repeating the process, is mostly manual – though some standards are emerging for managing it. Automation and management of production tasks, including deployment of the model, monitoring of its accuracy and periodic retraining of it, is likewise in its early stages, with tools just now emerging. These processes are bespoke efforts, which lead to inefficiencies, which lead to cost.

Great Expectations
Another cautionary note is that, to state the obvious, AI isn’t magic! That hyperbolic statement distills down to this more empirical one: there’s a vast discrepancy between what AI can do today and what executive expectations of its capabilities may be. You’ll need to manage those expectations. Failing to do so could result in perceived project failure, even when the outcomes are successful.

Beyond the perception of success or failure, managing expectations is also crucial to adequate budgeting for AI initiatives. We’ve already discussed the high costs involved in AI pursuits. Without sufficient funding, you’re setting the project up for failure.

The Bright Side
With so many caveats and cautions, you may get the impression that I’m bearish on AI, but nothing could be further from the truth. AI, if implemented properly, can deliver significant business advantage, today. What I’m cautious about – a caution I’m trying to you convey here – is that such proper implementation can be elusive. The answer to this quandary is that enterprises should avoid full AI implementation on their own, unless significant technology R&D is part of the normal operational procedure and budget.

The best way to benefit from AI innovation, and mitigate the risks of expensive and/or unsuccessful implementations, is to buy products that have embedded AI functionality. This shifts the burdens and risks of data science to the developer of the product, for whom the AI investment is quite reasonable, as it benefits numerous organizations (the developer’s customers) and not just one. Further expenditures, for testing, implementation and other engineering costs, are similarly leveraged and amortized.

Better yet, the benefits will keep growing. AI is getting better and the engineering process around it is slowly improving as well. That means products with embedded AI will continually gain more capabilities, and those capabilities will be slipstreamed into products already in use and processes already in place. That’s all upside, and incurs little, if any, disruption.

Staged Adoption
Eventually, AI engineering will become much more sophisticated and integration of AI into line of business software developed by in-house enterprise developers will become much more feasible. That will be a great time to integrate AI into your team’s stable of technologies. You’ll benefit from the added efficiencies and, by using products with embedded AI today, you won’t miss out on the competitive advantage of using AI early on.

As with any technology, AI brings with it a set of risks and rewards. Even if you are ready to take on some of those risks now, having an inventory of them can help. And, in general, if there’s a way to insulate yourself from those risks while exploiting the rewards as much as possible, you’ll come out on top. Carry out the necessary due diligence for your organization and see how closely you can stick to that plan, while still staying competitive and innovative.

About the author

Oksana Sokolovsky is an ex Wall Street executive turned entrepreneur; an experienced CEO who achieved early stage acquisition. Sokolovsky is passionate about developing disruptive technology. Her technology expertise combined with business acumen, allows her to bring a unique perspective to developing innovative products, commercializing them, and taking them to market. She is a technologist with experience running large IT departments within leading global Financial Services firms, establishing and transforming technology functions, and leading global high performing teams. During her 20-year technology career, Sokolovsky has held a number of senior roles at JPMorgan Chase, Morgan Stanley, and Deutsche Bank, as well as United Health Care, Instinet, and Barnes and Noble.com. Most recently, Sokolovsky built disruptive data discovery technology, which was acquired by Centrica's Io-Tahoe. Rohit Mahajan is an ex Wall Street executive turned entrepreneur. Mahajan is passionate about developing disruptive technology for data discovery using machine learning. He is an experienced technologist with a proven track record of implementing global solutions at financial institutions for devops, testing, security and data center transformation. He has developed technical solutions for global implementation and led major technology initiatives to drive transformation. In his 20 year technology career, Mahajan has held a number of senior roles at Dun and Bradstreet, Morgan Stanley, and Deutsche Bank. Most recently, Mahajan built disruptive data discovery technology, which was acquired by Centrica's Io-Tahoe.

You might also like...

Data Governance Operationalization: The Team

Read More →