- In this lecture we look at process applications. These are applications that use cognitive technologies to automate tasks or business processes internal to an organization. An example I love is the Hong Kong subway system which carries five million passengers a day with a 99.9% on-time record. One of the reason they perform so well is because of their program of preventive maintenance. 10,000 people perform some 2600 engineering works across the system each week. Scheduling all of these projects was a full-time job for the system's senior-most engineers.
They've developed an algorithm that automatically creates an optimal engineering schedule, saving them two days of planning work per week and allowing the engineers to spend time on other priorities. Another example of a process application is the state of Georgia Campaign Finance Commission. This organization processes up to 40,000 pages of campaign finance disclosures per month, many of which are handwritten. Faced with a deadline to make this data public, the commission deployed an automated handwriting recognition system coupled with human oversight to ensure quality and to enable them to scale up and handle the volume.
A third example is the Cincinnati Children's Hospital Medical Center, which like many medical research institutes, conducts the labor-intensive task of screening patients for eligibility for clinical trials. This hospital developed a natural language processing system to read free-form clinical notes and use machine learning to refine the list of terms extracted from those notes to automatically suggest patients that might be eligible for clinical trials. It reduced the workload of the nurses by 92%. The benefits of process applications are getting work done faster, cheaper, better, or some combination of these.
And these examples highlight common scenarios that we see in process applications. One is automating expert decisions. The Hong Kong subway example shows how decision-making and planning that require lots of expertise can be automated, not to replace but to empower the senior engineers who still review, approve, and sometimes modify the plans that the system makes. Another scenario is relieving skilled workers of unskilled tasks. In the case of the clinical trials example, the nurses which had spent so much time reading the clinical notes no longer had to do the reading.
They simply had to make the medical judgments based on it. And the final example is automating routine and unskilled work in order to scale up processes. That's what we saw in the state of Georgia example of processing written forms. So wrapping up, process applications automate tasks or business processes internal to an organization. We've seen examples that automate expert decision-making, relieve experts of unskilled tasks, and automate routine work.
- Artificial intelligence explained
- Cognitive technologies explained
- Supervised, unsupervised, and reinforcement learning
- Machine learning models and algorithms
- Language, speech, and visual processing
- Business applications of cognitive tech
- The impact of cognitive technologies at work
- Future of cognitive technologies