top of page
Search

The Future’s Ace: Diagnosing Problems with AI


Artificial Intelligence, otherwise known as AI, is a daunting term for many people. In fact, the condensed terminology “AI” itself appears to be a way for the average reader or layman to encapsulate and manage what might otherwise be an alien concept – the stuff of science fiction novels and doomsday prophecies. Though prominent figures such as engineer Elon Musk have raised concerns about AI heralding the end of humankind itself[1] and others lament the foreseeable obsoletion of their skills (with a mere 26% of retail staff supporting the implementation of AI in a recent KPMG study[2]), the modern day sees AI as a useful tool with plentiful applications in both social and economic hemispheres.

This article hopes to introduce the benefits of AI in several use cases; from applications in healthcare and finance to the identification of process problems, AI proves its mettle in optimizing and responding to a plethora of real-world issues. However, the technology does not come without its limitations. The use of AI introduces the need for new ethical considerations to determine the extent to which AI can be used, and by whom – not to mention the legal ramifications of using AI in a way that is deemed to transgress human rights.

Can AI Fix It?

AI is, essentially, the training of a computer to “replicate or simulate human intelligence in machines”[3]. With such training, the healthcare industry has seen a flourishing of healthcare AI, with AI used to improve disease diagnosis accuracy in conjunction with computer vision. The “stethoscope of the future” by one of SGInnovate’s startups uses such AI and computer vision to see into the body, with the AI having being trained with millions of images and subsequently learning what to look for in its analyses[4]. With such AI solutions that enable doctors and nurses to do their jobs more efficiently, limited resources in the healthcare industry can be dedicated to other areas requiring their attention.

In retail, where the forces of demand and supply are at play, AI can improve the process of sales by helping to predict demand and optimise inventory to reduce wastage. Historical data sets that are available regarding sales periods can be used to train it to expect when demand will be heavier – whether it be according to promotions or holidays – and optimise inventory to account for such fluctuations. Not only will there be a reduced amount of spoilage; retailers can know what to expect and consequently create schedules with a better work-life balance[5].

As for the justice system, AI can help alleviate the workload in areas that require plentiful manpower that cannot always be provided. In America, the Intelligence Advanced Research Projects Activity’s Janus computer-vision project allows analysts to perform trials in which algorithms are used to mimic the work of a human analyst, and distinguish one face from another. Likewise, the U.S. Department of Transportation is using AI to help in the detection of traffic accidents using video to improve commuting safety and experience[6].

In more technical areas such as engineering, present solutions to diagnose problems in processes involve the use of automation. However, this is not always efficient – where abnormalities should be raised as red flags for operators to identify problems before disruptions, the automation system usually shows operators many variables and leaves them to do the analyses. With artificial intelligence, the AI itself can accurately pinpoint the specific sensors that require attention, saving time and effort[7]. This allows operator to more effectively address problems and prevent outages.

Can We Handle It?

The limitations of AI are its cost of implementation – new technologies often come with high costs, especially if these solutions are bespoke. WebFX estimates that custom AI solutions can cost anywhere from $6000 to $300,000 a year, while third-party AI software can cost up to $40,000 a year[8]. Other limitations include the fact that the intelligence is limited to the data sets that it is trained with. As David Parmenter, head of data science at Adobe states: the key is to have “noteworthy data that will enable them to learn, that is appropriate for whatever task they have as a main priority”[9]. On the socioeconomic front, people’s resistance to change may limit the introduction of AI on a systemic level, while ethical and judicial frameworks also have to be laid out or considered before unsavoury, unprecedented incidents occur with the use of AI.

AI may very well be the key to the future, though some might argue that such expectations are too optimistic for the speed of our current technological development. What is apparent is that many companies and individuals are implementing AI in many current solutions that tackle problems more efficiently, and effectively than automation or manpower in several fields – proving the potential of AI to improve our quality of life when harnessed well.

References:

[1] https://www.livescience.com/48481-elon-musk-artificial-intelligence-threat.html [2] https://retailwire.com/discussion/why-do-so-many-people-working-in-retail-find-ai-scary/ [3] https://builtin.com/artificial-intelligence [4] https://govinsider.asia/inclusive-gov/singapores-healthcare-revolution-ai-key/ [5] https://retailwire.com/discussion/why-do-so-many-people-working-in-retail-find-ai-scary/ [6] https://www.ncjrs.gov/pdffiles1/nij/252038.pdf [7] https://www.controleng.com/articles/artificial-intelligence-identifies-process-abnormality-causes/ [8] https://www.webfx.com/internet-marketing/ai-pricing.html#:~:text=In%20comparison%2C%20custom%20AI%20solutions,%24200%20to%20%24350%20per%20hour. [9] https://www.analyticsinsight.net/top-5-limitations-artificial-intelligence/

37 views0 comments

Comments


bottom of page