Those two letters pack quite the punch when you consider how often you've heard that acronym thrown around lately in the media. If you're like us, then you're probably wondering what the big deal is and how AI will impact your life. Maybe you've even jumped on the ChatGPT train and had it write that cover letter you've been putting off. To help better our own understanding of the wonderful new world of AI automation, we sat down with members of our development team at DiscoverTec to discuss how AI is disrupting different industries and what implications it has for both our company and our clients in the healthcare and automotive industries.
Before we get too deep, let's sort out a few of the logistics. Namely, what is AI and how it works.
Artificial Intelligence (AI) is a term used to refer to machines that can simulate intelligence, meaning they can learn, think, and act in ways that are similar to humans. AI systems can make decisions, solve problems, and communicate with people in a way that is very lifelike. This intelligence is achieved through machine learning, which is the process of teaching computers how to think and act like humans by making connections between data points.
"[Machine learning is] basically teaching the machine how to predict something from something it already knows." - Charles B., Senior Software Developer
Senior software developer Charles explained how ChatGPT uses the same transformer architecture Google uses to translate languages. The system uses that transformer to analyze text and predict what the next text should be. For example, if you ask ChatGPT how much a car is, it will look back at all the information it consumed during its training and predict what words should come next; however, it doesn't have any understanding of what the answer should actually be.
AI will revolutionize healthcare for both humans and animals alike by providing better diagnoses and treatment options. Our team has developed a dog activity collar for a client that uses AI to detect how much physical activity the dog is getting throughout the day. The collar tracks movement and provides feedback about how active the dog is based on its behavior. When developing the technology in the collar, our development team used machine learning to teach the chip in the collar how to recognize different activity types (i.e. the difference between a dog walking and running full speed).
"This is what it looks like when a dog walks. This is what it looks like when a dog runs. This is what it looks like when a dog is resting, and so we teach it what these different activities are ... then when we put the collar on a dog, since we've taught {these} activities, it can recognize those." - William J., Software Developer
Software developer William goes on to say how the machine learning aspect of the technology will allow the collar to recognize an activity type it hasn't seen before and be able to deduce that it fits "into the paradigm of a walk." Essentially, the use of machine learning over a traditional algorithm is preferable because of its predictive capabilities.
This aspect is advantageous in the veterinary space because now your dog's vet can look at the collection of activity from your dog's collar and use it to inform their diagnoses and medical advice. For instance, if you have an active breed such as a border collie, and their collar detects something out of the ordinary like a prolonged period of rest, then this can help the vet see that something might be wrong and need treatment. Having that historical data and seeing how it trends has life-saving potential.
A key difference between the dog collar AI technology and a system like ChatGPT is the way each learns. Right now, our dev team must continuously update the initial data set the collar learns from—meaning any new activity we want it to recognize, it must be trained for. Conversely, ChatGPT has its initial data set, but its algorithm is actively learning in real-time, improving from its active users, and using those insights to train itself. Having this ability to teach itself without constant updating is the eventual goal for our dog collar.
"You know, just like any training, even training humans, ... if you train it wrong, it's going to get the wrong answer. So, you know, part of the challenge with training this is making sure you're training it with accurate information." – William J., Software Developer
Continuing in the realm of healthcare, DiscoverTec is currently working with a client based out of Texas that is an HR employee assistance program (EAP). This client goes to other companies and sells its EAP services as a white label. When asked about AI applications to this HR company's business model, our devs suggested some sort of chat service, like how ChatGPT operates, where the user could ask a question and be supplied an answer without having to dig through all the documentation on the site. The AI would have already mined all this information and be able to use a similar algorithm to ChatGPT to respond to inquiries and even suggest different courses of action the user can take.
“[The AI] could theoretically ingest all these help documents on the site, and you could set up a chat thing that someone could go ask a question in the chat box and get results based off of documents in that site.” - Robert G., Software Developer
For example, if the current EAP site is using quizzes to gauge a user's mood or feelings and then suggesting techniques like mindfulness or a specific diet to help alleviate any negative feelings, an AI chat service similar to ChatGPT could also provide answers and options for the user. But this avenue would make the process more humanlike and would emulate an actual conversation, lending to a more authentic and natural exchange. For something as serious as mental health, small improvements like this could have astounding effects.
The most important factor to consider when talking about the future of AI in healthcare is the matter of data security. Medical care is subject to several patient privacy laws that developers must be mindful of. The AI would be fed data that includes sensitive patient information and any breach of said data or any inaccurate outputs from the AI could pose legal consequences for the healthcare companies, providers, and the developers behind the AI algorithms. Microsoft is only allowing healthcare use cases to be used with its AI technology in a limited capacity under very strict protocols because of these implications.
This is not to say that AI won't play an integral role in the future of healthcare, but rather, the industry itself poses unique obstacles that must be handled delicately to protect people's health and wellness. We're talking about human lives here, which of course, should be handled with much greater care and consideration. For now, the law will have to catch up to the technology before AI can be fully utilized to its maximum potential.
“[AI] is an aide to help you do stuff, but it’s not to the point where it’s going to take over what you’re doing.” - Brendan S., Software Developer
Our DiscoverTec developers feel this further demonstrates why AI will never replace all human actors in processes; we need that final human check on the data and the programming to ensure the AI doesn't produce inaccurate and potentially harmful or misleading results.
"I don't think any of this stuff is ready for autopilot. If we're talking about ... the liability of a doctor using it or something, they should still be liable to check if you got it wrong. ... I don't think it would be replacing anybody. It'll just help us find some efficiencies in the work we do, but we still have to be on the hook for checking out that work. You have to peer review it." - Charles B., Senior Software Developer
DiscoverTec previously worked with a client that used scanning technology to troubleshoot and diagnose automotive issues in vehicles. When asked about incorporating AI into this process, our development team agreed AI could be utilized for preventative maintenance in cars. Basically, the AI could be directly in the vehicle and use data from scans, how the driver performs personally, and the car manual information to suggest when certain services might be beneficial. For example, if a driver were to put some extra miles and wear and tear on their car, the AI system could suggest when to replace the tires or brakes because it would be able to predict when these parts might fail based on the data it pulls.
AI also has many use cases for IoT, or the Internet of Things. IoT refers to the ever-growing network of connected devices that can send and receive data. AI could be used to monitor this data and make predictions as to what might happen in the future. When it comes to home automation, AI could detect any changes or anomalies in the data and then stimulate an immediate response from another device. For example, if you have a smart alarm system, AI could detect potential intruders and set off the alarm before they even get close to your home. At DiscoverTec, we're working with a client where we could utilize AI to monitor the inventory of a liquid tank company. The AI could help us find trends to help our client forecast what their inventory will be over time based on specific demand and other determining factors. Depending on the device, AI is applicable in many ways to IoT.
We've covered quite the gamut of real-world applications of AI automation. From tracking your dog's heart rate to answering health questions and scheduling your next oil change, the possibilities are both endless and exciting. As with any newer technology, there are some challenges and risks associated, but if done responsibly and effectively managed, AI could help us create much more efficient processes and systems for our businesses.
"I think ChatGPT has made [AI] very accessible, and they've marketed it in a wonderful way so it's attainable to the average Joe." - Brant W., Software Developer
The newer iterations of AI automation like ChatGPT have made AI accessible to nontechnical people with its use of natural English and its conversational chat interface. This brings new people and ideas to the table as far as what AI can be used for and opens the doors to new industries where AI might've previously been overlooked.
At DiscoverTec, we believe that AI is here to stay and will be a major player in how we operate businesses going forward. With proper data management and governance, the possibilities are endless for what AI can do and how it can help us become more efficient and productive. We aim to empower our clients with the knowledge they need to make smart, safe decisions when it comes to AI automation and help them stay ahead of the curve.
To find out more about how we're opening the door to AI at DiscoverTec, give us a call at 904-680-3000.
Published on: March 16, 2023 by Karli Walleser, Marketer