Audacious automation predictions and lists—they capture our attention, but how valuable are they to our business success? We examine how our predicted 6 trends have shaped capture and recognition:
- Deep learning will move from the honeymoon phase to the more practical.
While the media continues to be taken with all things artificial intelligence (AI)-related, and vendors still pitch solutions based upon machine learning, the enthusiasm for these solutions far outweighs what we see in terms of practical, innovative applications for businesses. We are starting to experience more in-depth analysis of applied AI in the form of articles that attempt to dispel the myths and false claims of vendors.
- Quality input data will be invaluable. Brokerages will pop-up to satisfy this need when the ability to independently gather a sufficient level of data is not practical.
Since the emergence of crowdsourcing services, such as Crowdflower and CloudFactory, we’ve seen these services that focus on providing clean data to businesses and solution providers—which need data to train their AI-based systems—gain a real foothold in the industry.
- Machine learning will not be an off-the-shelf-product, despite any vendor’s claims. Retail-level AI doesn’t yet exist.
To date, no advanced AI-based system provides a machine learning platform out-of-the-box. Most of the attention has focused on the lack of experienced staff to work with these platforms and the resulting need for increases in compensation.
- The Law will play catch up with AI. A sunshine law of sorts will become part of the conversation regarding AI and machine learning as more decisions are left to the artificial minds of computers. This will help ensure scrutiny of how models are developed and avoid inherent biases.
The government has yet to step in, but we have seen an increase in calls-to-action by industry luminaries, as well as recent inquiries held on Capitol Hill regarding the ethical use of AI (some driven by alleged election tampering). These will undoubtedly strengthen the debate and fuel calls for real legislation.
- Data-as-a-Service models will grow and multiply, as we’ve seen with the emerging algorithm markets. Traditional businesses (apart from the big data companies) with treasure troves of “actionable data” will explore how best to monetize their data lakes and begin the first steps toward this goal. They will also make critical strides in replacing their “big data” analytics with focused, rapid and actionable data analytics so that more of the “right” analytics questions get asked and answered.
It’s hard to quantify how far companies have gone in creating a revenue stream for data that they already curate within their business processes. The visible signs are available with the usual suspects such as Amazon, Google and Facebook as well as companies such as Comcast and ATT, but there are many, many other businesses that actively collect data. How and under what circumstances they use this data as revenue is still a mystery.
- API-as-a-Service in cloud computing (both public and private) with machine learning underpinnings will continue as a major trend where the underlying technical details are abstracted away from the businesses that use them. Few installation steps and no maintenance will be required. Businesses that succeed in developing their cloud strategies will link their IT services to business outcomes and see significant gains over their competitors in 2017.
Thousands of API-as-a-services exist, but the level of true machine learning that underpins them is varies dramatically. Even services such as IBM Watson that use deep learning are still narrow in scope.
Three 2018/19 Trends: Automation and the Role of AI
Let’s explore the future of automation and the role artificial intelligence will play in it in 2018 and 2019. Our predicted trends for this year and next build upon what we observed from this past year:
- Bridled enthusiasm increases for AI and wanes significantly—depending on the application.
Confusion over just exactly what level of AI is involved within software solutions continues in the marketplace, but more businesses require clarification and demand to know how to actually benefit from it. Those expecting AI to advance to the level of simply replacing humans by offering more general capabilities will be disappointed while those that are targeting very-specific areas to reduce effort (e.g., assistive applications) will benefit further from applying machine learning techniques to current work. Specifically for capture software, the ability to dramatically simplify the initial configuration and ongoing operations through machine learning automation will be essential. We will begin to see the additional steps toward adopting capture powered by machine learning in 2018.
- Cloud capture increases adoption for specific applications.
Largely driven through more machine learning capabilities that are adaptive, cloud capture (e.g., either multi-tenant or API-as-a-service) will experience continued traction. This prediction is predicated upon our #1 prediction since so much of the configuration or ongoing performance must be offloaded to advanced software without requiring novices to become experts in document capture.
- Robotic Process Automation (RPA) is put in its place relative to other technology domains.
While analyst firms such as Horses for Sources (HfS) and Everest Group have solid definitions of RPA relative to other technologies including analytics and “cognitive computing”, the actual users of this technology don’t necessarily distinguish between RPA and] more-advanced AI-based systems. RPA, in this context, is an easier-to-configure rules-based expert system that automates very-specific tasks. RPA, in many ways, is the lowest level of the automation continuum and, therefore, the least “intelligent.” While RPA has become a popular buzzword with a TLA, the limits of this technology will become apparent in 2018. For those unsatisfied by RPA, they will be looking to compliment this capability with more advanced AI-based technology.
For 2018, there’s clearly plenty to be excited about with many opportunities and some equally daunting challenges. Get ready—we’re in for a wild ride.