Subscribe
Share
Share
Embed
Dr. Hal King is back this week to discuss the crucial role of technology in ensuring food safety. Technology has many roles to play in food safety, from validating food certifications to helping businesses establish effective food safety cultures by taking out the guesswork. Dr. King emphasizes the value of artificial intelligence and machine learning for collecting raw data and converting it into actionable information. Dr.King and host Dr. Darin Detwiler also discuss the dangers of over-reliance on technology and the importance of human oversight and intervention. Despite some pitfalls, technology is a powerful tool for establishing and maintaining trust and preventing foodborne illness. Listen and find out how.
Creators & Guests
Host
What is Trust Bites?
Hosted by food safety industry leader and consumer advocate Dr. Darin Detwiler, "Trust Bites" examines the challenges of ensuring food safety in a complex global marketplace and maintaining brand reputation.
With the rise of global food trade, consumers, retailers, and producers alike are increasingly concerned about the safety and quality of the food they buy. Many existing validation systems are outdated, bureaucratic, and expensive, creating inefficiency and allowing vital information to slip through the cracks.
"Trust Bites" delves into the validation process, discussing the challenges and limitations of current systems and exploring fair and equitable solutions.
Dr. Darin Detwiler: Hello, welcome to another episode of Trust Bites. I'm Dr. Darin Detwiler, your host, and I'm very happy to be joined by one of our advisory board members, Dr. Hal King.
Hal King: Thank you, Darin. Happy to be here.
Dr. Darin Detwiler: Tell us just a little bit about your background so audience has an idea who you are.
Hal King: Yeah, Darin. I run an advisory services firm that helps the food industry, primarily the food service industry, with multi business locations around the country and around the world, to prevent foodborne illnesses. We really work to make sure that those businesses have food safety management systems and programs in place and can achieve food safety culture through management and business leadership. So we don't really take projects that are quality-specific, we're just all about trying to help these businesses do the right thing and put better systems in place to prevent foodborne illnesses.
Dr. Darin Detwiler: I've got to add, your years of service in terms of your work with the CDC and the public health service and even in military in terms of food safety has definitely not only supported what you're doing now, but have supported entire industries and businesses and consumer safety for many, many years.
Hal, you're not alone in doing this work. There's a good number of people who you could point at in terms of the, well, I refer to it as the Herculean effort that's put into food safety and, you know, having been asked to define Herculean effort, I choose three words: work, strength and courage. And while I thank you for your courage, I want to point out that helping you with your strength and the work you do, technology has got to play a role in it.
Let's focus on technology. What are your thoughts on the role of technology in ensuring, whether it's transparency or traceability or validation, but the role of transparency in terms of protecting public health in the food industry?
Hal King: Thank you, Darin, for that question. It's a really important question in our industry today because a lot of people are just starting to use technology and if they don't use technology the right way, then it doesn't really provide them the benefits it can.
When I first went into the restaurant industry business, I think in 2004, you know, most everybody was using paper checklists and running around the restaurant checking temperatures. And when I observed that, I was kind of shocked. I was like, "Well, who can see these checklists? They're on pads that no one ever really goes and looks at. Someone does it and didn't even have at the time ways to correct issues that they might've found. Like, there was no definition of what the temperature should be. It was just a checklist with, "Here's a check, take the temperature of these, you know, this product." And so obviously there's a great example of... it's kind of useless.
Technology can be used in those situations to inform the employee or the manager who's going to do the test what he should be checking, what's the high hazard? When does it need to be checked? What are the specifications? What's the temperature supposed to be? And then when it's not the right temperature, what should they do with that food? Recook it? Or toss it or talk to the manager?
And then also, have a process in place to ensure they don't just skip that and walk away. "Well, we're at the right temperature, but I'm too busy."
Technology can ensure that that food is safe if it's used properly and it's designed properly to put the kind of checks and balances on following through and kicking that information.
You know, the bigger picture is, if a restaurant business has a thousand restaurants, how do they know that they're cooking the chicken every day properly in that restaurant? Or hamburgers, or whatever? Technology can then take that information and provide it in a format that they can see every single time that's checked. And they can see every single time it was corrected. They actually can have systems with technology to tell them and notify them that it wasn't done. And it could notify a responsible party in the restaurant to go back and check it. Almost like, instead of waiting for people to run a stop sign and hoping a policeman's there the day that they run the stop sign, every time you run the stop sign, the camera captures your license plate and you get a text: "You just ran a stop sign, you're going to get a fine." I mean, which do you think is going to have the biggest impact on compliance? I think the one that people know is going to hold them accountable. People are going to have observations of that data and look at that data.
So technology is really, really critical because it can take us away from papers, systems, and things that people don't really have a way to manage and ensure are taking place and give us tools to be able to know that's happening.
Dr. Darin Detwiler: I like how you frame that, because I would imagine that you and I have some observations, maybe even some photos we've taken our cell phone or some emails that we've sent off that is like: the average person would not see this and rattle off an email about this observation kind of thing.
But I was recently at the airport and I saw that there was a gentleman loading sandwiches and salads and yogurts and hard boiled eggs and, you know, very perishable items to people walking by. That's the first thing they would see. Refrigerated series of shelves, kind of a thing display. And the very first thing I noticed is that the the analog thermometer there was in the red and that the digital said that it was at 49 degrees.
And I pointed this out. "Why are you loading this here? Are you aware of this temperature?"
And I heard, "Well, yeah, we write down the temperatures on a daily basis." He said daily, I don't know if he meant hourly or whatever. And then come to find out that, "Oh, we sent in an order about a week ago or two that someone needs to come and look at this refrigerator unit."
So I fly out to New Jersey as it was, I fly back to LA a couple of days later and guess what? It's still in the red and it's-- I don't remember what the temperature was. I think it was 46 at this point, and I talked to the manager and the manager verified: yes, they did write up something about this a week or two ago. And I was also told that, "But you understand we are required to sell this, right?"
Well, wait a minute. Hold on. You're required to sell it, but you can't get it out below the danger zone. So, I go through the steps of finding out who is in charge of this with this company, and I rattle off an email. Again, I saw this Monday of last week as we're recording this.
Today I got an email from this person saying, "We will look into this and I promise that someone has their eyes on it within the next week." It's like, wait, but when I was in New Jersey, I'm talking with a company that says that they're using technology such that they have a regional manager that if there's any refrigerated unit that it's outside of their temperature range, not only does it alert them locally, but it alerts them at a regional level, so they can be aware that they have you know, an issue with a temperature within their region.
To me, look at the difference between technology giving instantaneous notification, alert, whatever to not only local but regional level leadership versus a paper trail of promises and emails and "we'll get to you" and "now it'll be three weeks at least."
But originally when I talked to him, they said a week or two. This could be at least five weeks of out-of-temperature compliance. At an airport for this perishable food. That to me is one of those, you know, incredible examples of just the difference between a paper based system and technology.
Hal King: You know, Darin, there's a great example of this same kind of situation that kind of affects a, let's say a restaurant and a vendor that's supposed to be taking care of the restaurant. Without the use of technology it's not going to change anything. I don't know if you remember years ago, there was a restaurant business in New York City that... People were filming rats running around the restaurant while it was closed, and they're doing it with their phone. I've been in that business long enough to know that sometimes when you're in New York City or Chicago or Florida, you're going to have rodent infestations. So you have to find ways to prevent them, right?
One of the most common things that happen when rodents get into an environment and they're there at night running around on surfaces and stuff. And it's just, it's horrible. It's unhealthy. It causes foodborne illnesses. They carry salmonella and everything else, you know, all over the place and they urinate everywhere.
This sounds gross, right? And one of the ways you can get confidence that a restaurant can operate safely under those conditions is first, obviously, get the rats out, right? And exclude them and put things in place, clean up everything, double sanitize everything, you know, make sure the environment's clean. But when you're in a high pressure environment, like New York City, other places, they're going to come back.
And so one of the criteria in those years ago was, we'll have our third party pest control come in and inspect the traps to tell us if the rats are back or the mice are back. And so, obviously coming back every few months or every 30 days is not enough, because first of all, they may see them, but it's a little bit late. How long have they been in there? And now they're back and testing the restaurant and oh, they now found the traps are full. You don't want the employees to go check the traps are making food. We don't need that for that happen.
So how do we get a business that come third party, come in and inspect traps every day when, when we really need that at a minimum to ensure that we're preventing reinfestation of that restaurant? They're going to charge us $150 every time they come out there and they don't even have the personnel to come in every day. They've got hundreds and hundreds of restaurants in every city and every locale where they're actually servicing these different brands.
Well, how do you do that? Well, technology allows you to put systems in traps. And as soon as they capture a rat or rodent, it can notify the manager. And so if a person like the vendor, third party can't get out there and remove it, you could shut the restaurant down. You could protect the public health by knowing that you're getting suppression on this issue, or the vendor can actually come and the restaurant shuts down temporarily and they get rid of the rats again.
Technology is moving us to a place where we can get more rapid ways of corrective actions. Otherwise we just leave them there, and just like you said with temperature, I would love to see us have a temperature monitor on hot water. How many times have I been in restaurants and I go, "Where's your hot water?"
"Oh, the hot water tank busted or broke."
"When did that happen?"
"Oh, a couple of months ago."
Well, you know, if the health department goes and finds that there's no hot water, they would shut them down, but they can't get in there every month when these events happen. But if a monitor system was in place, it says there's no hot water, we need to get action on this.
And of course, I think we talked about this in the first episode about trust: you are going to have to have companies with integrity and put trust in place to ensure that that's being done right. But we do need the systems and technologies that help us do that. Because those that need to do it and should do it, maybe one day there's a regulation for it. Hey, if you're going to run a restaurant, we need to see your data on hot water temps from your monitoring technology. I hope that's the case one day, because those are the ways we can have a regulatory authority check technology to ensure food safety, not just inspecting restaurants.
I'm not saying inspection of restaurants is, is not a good thing. It's important. But we need to have them link to the technology that helps us have better assurance of safety in the restaurant.
Dr. Darin Detwiler: Well, you know, insurances of safety and restaurant retail, it's not limited. We've talked about temperature control. We've talked about rodent control, but, you know, there's a lot of conversation, especially at the FDA level, about transparency and traceability. And when we look at, whether it is, you know, can we verify that this came from this location? Can we verify that it was produced in an allergen-free location? Can we verify that it actually met the requirements for being certified as this or certified as that? There's so much effort that goes into that certification. And yes, there are incidents of, and examples of, of counterfeiting or falsifying information, but if we just simply rely on the one step back approach to, "Well, I got it from my distributor," or "I got it from this person, but I don't know where they got it from," again, that trust, being engaged in a part of that trust, I could see how technology could help in that transparency and traceability of those very quality certification, validation efforts of your foods.
Are you seeing this now with the pressure on restaurants to be able to, you know, perhaps use technology to validate the certifications of the food that they say, if it is allergen free, or if it's, you know, whatever it is on the menu that we're not just taking the word from our distributor, we're going beyond that?
Hal King: We all know about the FISMA requirements for traceability and restaurants are going to be required to comply to that, be able to show an FDA investigator where the food came from in real time in that restaurant in that day. And that's good for FDA and CDC because that helps them kind of eliminate where the source might not be and find where the source of the contamination that causes an illness or during an outbreak investigation is going on.
But I'll tell you where the businesses are seeing the value of this. Now they're seeing the value within their own internal product withdrawals and recalls. So here's how it works. Technology's there. They can look on their system and go, " We get romaine lettuce from three suppliers in a thousand restaurants in 25 states. And we just got told by a romaine supplier that they had the USDA come in and find stuff in one of the processing plants and they want them to consider voluntarily pulling this product." Not a recall, not national, not reported, no foodborne illnesses, right? We want you to consider. Well, trustworthy companies would want to try to find where that product is, okay, and pull it out. They don't want to go and just pull out all romaine lettuce in all restaurants, because that's a huge cost.
Well, now they have a tool. They can say, "You know what? From that lot and that date and that processing plant, I could actually find the needle in the haystack and I only had to pull it out of 50 restaurants and I can tell them, notify them, pull it out and I don't serve it."
I'd basically just help prevent a potential, you know, issue with a foodborne illness because I was able to control that and no one would know about it. No one.
Dr. Darin Detwiler: It wasn't just in terms of foodborne illness issue. You saved, in this scenario, significant economic impact for the company. You're also talking about wasted food and, you know, access. But it's still trust, right? You should be able to do that and communicate that you're doing these things because you want your customers to trust you.
Hal King: A few years back I was traveling New York. We were all kind of working really hard, right close to Christmas. And the CDC and FDA said, "Don't eat romaine lettuce, no romaine lettuce in the whole United States, don't eat romaine lettuce." All my clients were calling me, "Well, what do we do?"
I said, "Well, you can't serve romaine lettuce because they just said it was contaminated."
They go, "Well, how do we know all of it's contaminated?"
I said, "We don't, but they don't know the source. And because people have died, we have no choice. We have to take it off the system."
We all know what the cost of that was. We all know the benefit. The benefit probably prevented illnesses and deaths, but the cost was huge. Could we have gotten better at that? Yes.
Here's what I was hearing from customers: I was talking to the CDC one night. I was in New York City, crossing the Brooklyn Bridge, and I was on the phone with CDC and one of my clients. I'm going, "Okay, we need to kind of figure out what we're going to tell customers, because customers are coming into these restaurants and saying, 'how do I know that you're not serving romaine lettuce?'" Because some restaurants were still trying to serve romaine lettuce, because if you remember during the time of the outbreak, the CDC came back and the FDA came back a few days later and said, "If you can show the romaine lettuce is not from this area and you can show it's locally grown."
Well, what is locally grown? On my rooftop, on my restaurant, in my local gardens? If you can show it's locally grown, you can serve it. Well, then restaurants start serving romaine lettuce. Customers are coming in going, "How do I know that's not part of the romaine outbreak? How do I know that was grown locally?" Well, the restaurants had to show the customer the proof It was grown locally.
I hope one day that that also is a way or a tool customers can go in and ask the question you asked, Darin. "Hey, how do I know that this doesn't have peanut allergen in it?"
"Well, let us show you," and click, click, maybe it's on an app, click, click, this is what they're doing. "I have confidence there's no allergen in this food. I have confidence they're not serving recall product."
But I think early on, like I said, the biggest value, and I love this, is that there's an economic value on traceability now, that customers can limit the amount of food and food waste, and obviously food security in lots of situations, to just find the one that's most likely contaminated. Even if it's not part of an outbreak and it's been associated with a positive sample of listeria, we can take that out. Trustworthy companies will take that out of the system and limit the economic impact on their suppliers.
So it just puts all the good actors-- you know, there are bad actors out there-- but because the good actors that make this business value proposition work. And then over time, all the other businesses kind of see that and they'll continue to do that.
Dr. Darin Detwiler: Well, on that note, some of the actors, especially, I guess, behind the scenes with technology are not human, right? We're seeing increases in robotics, but the biggest increases is in the area of artificial intelligence and machine learning. The idea of, you know, maybe it doesn't take the eyes of someone to see that this has been above temperature for so long. Maybe we can program into the system the idea of looking at that pattern and looking at that tendency and saying, "Look, maybe if the data shows this over time, then it automatically sends out a report rather than someone having to put their eyes on that, and make those actions.
Let's talk about artificial intelligence and machine learning. Can we talk about benefits and then maybe talk about, are there some potential risks or hazards or even obstacles to overcome?
Hal King: Yeah, the biggest benefit to me is what you just described. I have hundreds and hundreds of suppliers and I'm tracking surveillance data on pulling product, looking for bones in my DC's because I got X rays of the chicken, say, in my plants. I'm monitoring my customers to see if they got a bone in nugget. Customers call in, or they write in, "I got a bone. I bit them to a bone." I've got this whole milieu of data.
I can't put my eyes on that. I can do exception reporting. Let's say I get an email that says you've got a bone, someone bit into a bone and broke a tooth, whatever. And so I get, that works. I got exception reporting, but that works if I have a restaurant that maybe has a hundred transactions a day.
If I got a thousand restaurants doing 2 million transactions a day, I can't put my eyes on that, you know, even with exception reporting. That was what we were doing in the old days, right? Getting the email saying, " You got a problem here," and now I can put my, got my folks on that. But there's so limited amount of resources to be able to do that kind of work.
AI gives us the ability. I mean, think about this: and I thought about this the other day, about these new AI models that are connected to the internet. They can feed all this stuff that's on the internet through the AI. And not that it's going to go read everything in there, but it will find what it needs and it has access to it. And it's quick. I've used one. I asked it some really tough questions. I said, "Who's Hal King?" He did this, something that no one would know this. I was, did research funded by the WHO with the CDC and, and the Carter Center in West Africa and Ghana in a city called Accra on a project. It was funded by the CDC.
So all those are public records, but you would never be able to find that out about me. This thing finds it. It talks about what was published, the papers that came out of it, the grant that was funded, the people that were responsible. If it's out there, it can find it. So think about the data that could actually be monitored to help us have better assurance of safety by an AI. Like you said, it's not having to put our eyes on, it's monitoring. And now exception reporting is, it's already done something about it. It's not just, okay, here's what we found, but it's taught to have action. "We found this, this is what we found." Now it actually takes the action. It calls the manager, checks the manager, gets confirmed. The data is still there and it comes back and gives a loop back and you may not ever see it. It's over here just saying it was taken care of.
Could be that way in suppliers, checking products from suppliers to the D.C.. before it gets to the stores. AI gives us the ability to check things faster. So while it's en route and transit from a distributor to a restaurant, you even have the potential of AI redirecting that product before it even gets to the restaurant.
So I just think AI has a huge amount of benefit to be able to help us get better at monitoring safety. I hope that maybe the FDA does it too. We need better systems and manufacturing plants where AI can tell FDA, "These are the plants that we should be inspecting because we don't have the resources to do that."
Frank Yiannis mentioned that before. We said they use AI systems to determine which ships and containers on ships they should inspect to try to find bad actors and foreign imports of food, right? And I think they did 300%, he said, 300% better at catching which ones they should go inspect based on risk models from AI.
So I think it has a huge potential to help us.
Dr. Darin Detwiler: Well, it's so easy to get caught up in the how behind doing things, right? How are we going to do this? How are we going to do this? If you're looking at technology, you know, one of the first complaints you hear is that, "Oh, we collected all this data." Okay. I get that. You want it to convert the big collection of data to actionable information.
Okay, so now you have this actionable information. Well, we actually, actually have to take action, right? Because you don't want to just sit on this actionable information. And what good does it do to take action after the fact, after consumers have already been harmed, after the food's already been consumed, right?
So that timely response to action, based on actionable information requires a wealth of data to be collected and a lot of thought into when and where are you collecting this data and what questions are you asking and things of that nature. So it's a bit of a process. It's not just collecting the data.
And the artificial intelligence and the machine learning can really help us fine tune that pyramid up to that most desirable, you know, we're taking action immediately and we're getting results now, because this is where you're going to see the value. In reliance on technology in this way. And also where you're going to be having your most options available to you and the most impact in terms of reducing liabilities, reducing, whether it's food waste or, you know, economic loss or obviously consumer harm.
So it's one thing to look at potential benefits of it. Let's flip it before we end our session here. What could be seen as some potential risks or threats or obstacles in terms of technology like this?
Hal King: We talked about this in the prior episode and talked a little bit about this this episode: misplaced trust. If we trust this so much that we don't need to validate it and verify it, you know, validating mean it's doing what it's supposed to be doing. Verifying meaning it did it today. Right. And so to me, if we over-trust this without those two steps on a regular basis, maybe we're using AI to help us verify it's working with the checks and balances. But the biggest risk is we over-rely on it. And then we let things get to the point based on volume and seeing, and then we have outbreaks.
We over-rely on the risk model because the risk model is only really as good as what we put into it in a risk model. AI and the machine learning gives us the ability for the machine to learn based on, if this, then that, if this, and that. You know, those are the old risk models. If this, then that. AI and machine learning says, if this, then that, then I do this. And now if this and that I did that now, what do I do? It's that ability for the machine, you know, the computer systems to learn what should be done, but if we don't have checks and balances on that, it may spin out and do the wrong thing based on what it appears to think is a risk. So, you know, obviously giving it too much control and too much less, you know, little less oversight, I think, could be dangerous. But if we took it slow and validated it and verified it over time and make sure those checks and balances in place, I think it could be very beneficial.
Dr. Darin Detwiler: It comes to my mind that there's a lot of movies out there, sci fi movies that that show exactly what you're talking about in terms of, if you just give your responsibility, give your trust over, give your surveillance and your transparency and traceability over to technology and let them do it and sit back and go, "I don't have to do anything," the technology becomes too big and uh, takes over the world or whatever. I'm thinking like some of those like Terminator and "I, Robot" kind of movies.
Hal King: I like to use this example of people and I'm not promoting any one brand over another, but you know, I worked at Emory University and WebMD was created by Emory University doctors and it's monitored by doctors, right? But if AI goes to the internet and tries to find information, how does it know what's credible? How do you teach it what's credible? And so credible data we know might be people that we trust and people that have degrees and people that have certifications. They have to take tests and and be board certified, right? But if a computer can get board certified, where do we find our trust?
And if they're going to the Internet to get all their information, I think it's going to be really key that we have that integrity of, wow. What is trustworthy? What does a machine learn and how does it act so that we don't have those situations where it's allowed to do it?
'cause when it just goes to the internet, you could find all kinds of things about a subject matter you're looking for. And they all sound really trustworthy and credible, but they're not, they're just people trying to sell something or people trying to sell you an idea or whatever.
Dr. Darin Detwiler: Well, maybe a good way to close out this episode is that there's a case out of Portland, Oregon that I was interacting with the family a while. This is, this is years ago. This is almost a decade ago or so.
Father was driving the four year old to see an urgent care doctor and basically got on a cell phone and called his brother and rattled off the symptoms, what he was observing.
And the brother did, like you say, went on a WebMD and came back and said she might have E. coli. And so the dad went into the doctor and what was found later was there was the word E. coli and a question mark on the intake notes, but the doctor didn't see enough evidence to warrant scheduling any tests to be taken, and basically said "You should follow up with your family doctor within the next week or so," and sent a four year old with bloody diarrhea and other symptoms of a food borne illness home.
That child was dead within two or three days because of the stage of renal failure that she was already in.
So even when you have times where the technology can tell you, "Based on what I'm seeing, this is the recommended course of action," we still, as humans, have to follow up on either taking those actions or validating, verifying that the actions were taken.
And if they were taken, did they achieve the actual desired goal, whether it's reducing rodents or lowering the temperature or allowing the true traceability of that certification or validation, from, from, from farmer to fork all the way through.
It's never going to be a scenario where artificial intelligence or machine learning truly eliminates our roles as humans. But it can definitely help us, again, with the work and strength that we have to put into it as long as we partner with our human courage to make sure that we're continuing to put a Herculean effort behind food safety.
Hal, thank you very, very much. And I look forward to talking to you in future episodes.
Hal King: Thank you, Darin. It was a pleasure.
Dr. Darin Detwiler: All right. For Trust Bites, presented by My Trusted Source, is Dr. Darin Detwiler. Thank you very much for joining us.