UNCC 3MT 2018

All right. Hello everyone and welcome to the University of North Carolina at Charlotte's 3MT competition, also known as 3 Minute Thesis. So my name is Miranda Rouse and I will be your master of ceremonies today. I'm adjunct faculty here at UNC Charlotte in the Communication Studies Department. I also am an associate in the University Speaking Center, which is located here in the library on the ground floor. And that's a free resource for all students to work one-on-one with a peer. I also work part-time at a nonprofit radio station, WGFY, as their fundraising and community coordinator. Also, I'm a formal— former 49er and also worked with the Center for Grad Life as a GLF. So speaking of the Center for Graduate Life, we would like to thank The Graduate School and the Center for Grad Life. So if we could just give a quick round of applause for them putting this event together. All right, I'm going to start by giving you a quick background of the 3MT competition itself, and then we will go into more details.

Okay.

All right.

So the first 3-Minute Thesis Competition, also known as 3MT, was held at the University of Queensland in Australia in 2008 with 160 research higher degree students competing. In 2009 and 2010, The 3MT competition was promoted to other Australian and New Zealand universities and then developed into a multinational event. So since 2011, the popularity of the competition has increased and grown, and the 3 competition— 3MT competition, excuse me, is now held in over 350 universities across over 18 countries worldwide.

Worldwide.

The 3-Minute Thesis Competition celebrates the exciting research conducted by graduate students. In 3 minutes or less, the participants must effectively explain their research in a language that is appropriate to non-specialist audiences. The preparation and presentation combines the academics with their research and communication skills. This competition gives students an opportunity to publicize their research and also show themselves as researchers. So now let's go ahead and meet the judges for our competition today. All right, first we have Shahar Akhour. She's a 2017 UNCC Graduate School alumna, and she works as Senior Industrial Organizational Analyst for the NBA. Next we have Cynthia Wolfe Johnson, who is a former UNCC Associate Provost for Academic Services here on campus. Then we have Michelle Harris, who is a training designer for Harris Teeter. We also have Karen Arrington, who is an Associate Director in the Center— in the Career Center. Excuse me. She will be filling in for Christy O'Connor today. And lastly, we have Lauren Shibbs, who is the baseball coach here at UNCC.

Hi, everyone.

All right, so now we're going to go over the rules for the 3MT competition. So each participant must present within 3 minutes or less with the use of one single presentation aid. The participants will be timed, and if they exceed the 3 minutes, then they are disqualified. Today, the competitors will be judged on 2 different criteria, and the first one is going to be on comprehension and content. And this means, did the participants follow a clear and logical sequence, and was the thesis, key results, research, and outcomes significant and communicated in a language appropriate to non-specialist audiences? The second part will be engagement and communication. And this includes, did the presenter capture and maintain the audience's attention? And did the PowerPoint enhance the presentation?

All right.

So now we're going to discuss the prizes for today. So there will be 3 prizes. The first being the People's Choice. Then we have runner-up. And we also have first place. The first place winner will receive travel funding to attend the regional competition at the Conference of Southern Graduate Schools in Knoxville, Tennessee this February. The judges will have 2 minutes to write comments following each presenter. And after each participant has presented, the judges will deliberate and tally all scores for the first place and runner-up winners. The People's Choice Award is going to be decided by you, the audience. Ballots will be handed out and collected and tallied while the judges are deliberating. So we encourage that you all stay till the very end so that you can cast your vote for that. All right. So we are going to go ahead and get started with our first competitor. Which will be Sara Abdellahi.

Hi everyone, my name is Sara Abdellahi. I'm from the Department of Computing and Information, and my research is about designing AIs which are able to read your emotions and use them. So I assume all of you have some experience with collaboration of some sort. Or at least observed people collaborating on tasks. I want to ask you to raise your hand if you care about how your collaborator understands your feelings and cares about the interactions they have with you and it's not all just about the task. Would you raise your hand if you care about that?

Yeah.

Yeah, good. I bet there is no one here who says, I actually like a crazy collaborator who doesn't care about my feelings and And there's someone like the Sheldon Cooper from the Big Bang Theory series who is not understanding of anyone, and that's the perfect person I want to work with. So when people collaborate with each other, they look into 2 main processes, unconsciously of course. One is the function, the task they're performing. The second one is the interaction they have with their collaborators. But for some reason, so far when we have been looking into AIs, we skipped one of the two. We only focused on the function and we tried to make our AIs smarter and smarter without thinking, okay, these AIs are going to supposedly work with human beings and they need to understand them. So we are designing basically annoying AIs. So what's the solution here? What I'm proposing for my research is actually the path science fiction is going. Just dislike the comedy where we get all those annoying characters. Science fiction is always great with suggesting very interesting ideas ahead of all of us. And one of the things they're very much focusing on is AIs with capability of reading emotions and responding to emotions. And that's what I'm presenting a model based on. An AI which can capture your emotions through different technologies existing out there like eye tracking, like different temperature capturing, and so many technologies we have, face recognition, all sorts of technologies, and then use that captured emotion to decide, okay, what's my collaborator's state and what's the next step or state I want my collaborator to move into and how I want to maintain that through changing the emotion. That's what I'm mostly focused on. Thank you for your attention.

All right, up next we will have Akshay Ayanchira.

Good afternoon, everyone. My name is Akshay. And I'm here to present my ongoing research topic, which is Augmenting Reality for Emergency Evacuation. So to start with, Mr. Tony Stark's image has nothing to do with my thesis, but I purposely kept it here to convey how fascinated we are watching sci-fi fantasy movies like Avengers, because it makes us think if something of this kind is doable or not. So one day, me, my professor, and my labmates were sitting in our lab thinking if we can use the technologies that we have to help save some lives. So we went forward with one of the biggest concerns, the fire hazard. We pulled up last year's stats from NFPA website, and these are some of the figures that we got. It says that more than 13 million calls were registered last year across US, out of which 5 lakh buildings actually caught fire. So 5 lakh buildings in a year is like saying, by the time my speech ends, we will have 3 buildings already catching fire. Yeah, and that's how bad the situation is. So, So to learn more, we called 5 officers from Charlotte Fire Department to our lab and asked them questions like what difficulties they face, where do things go wrong while performing a rescue operation. And they came up with 3 strong points. One is sometimes they ended up going to a wrong address, a wrong lane altogether. So that's a disaster. Second one is even if they find a building, it's difficult to locate where the people are inside a building. And third, when they get trapped inside, it becomes difficult for everyone to find an evacuation path. So to tackle this, we proposed a solution where a trapped user's smartphone can synchronize with HoloLens. HoloLens is this device which will be worn by a rescue coordinator, and when these devices sync with each other, rescue coordinator can not only detect where a building is in a particular area, but also precisely see where each user is moving inside a building. May it be a small house or a 4-floor university building. And we are not stopping there. We are keeping the software framework open for multiple types of input parameters. For example, this is the 4th floor of Woodward Hall that we modeled. In Woodward, we have many cabins and labs equipped with heat sensors. Now that is a good extra parameter we are working on. Using heat sensor information, we can identify which all areas are already affected by fire. And we can prevent users from going there. And using the same information, we can also predict the best possible evacuation path for everybody to come out safe as early as possible. Using this, like, the officers were impressed, and they want this kind of technology to be integrated as soon as possible in their helmets. And using this thesis, I believe I will be able to help our real-life Avengers, the real-life heroes, to a great extent.

Thank you.

Thank you.

Okay, I'm good. Yeah, yeah, yeah. You're fine.

You got excited. You got excited. All right, up next we will have Gizem Bouchak Zilar.

Yeah, thank you. Hello everyone. I'm interested in collective behaviors, especially in collective behavior, collective decision makings in our society. Think about it. You are walking on the streets.

Should I go right?

Should I go left? You know, we don't know. So every time, maybe we are making decisions collectively. But especially when it comes to for the same cause made us together, it could be for protests. This image is from after school shootings. Students came together and went out to show their reactions. Unfortunately, it was a sad. Of course, it was a sad emotion. But at the same time, they wanted to come and show their reactions on the street. What they did, some of them have their banners. They wanted to show, you know, with their banners, I'm young, I'm angry, and I'm ready for change right here, right now. But some others, they want— as you can see in this yellow squares, they are with their phones. They are texting, they are posting, or maybe they are tweeting. And outside of this arena, also there are other people who are talking about this issue. So who are they then? What they are talking about, we are looking for their social media data because they are sharing on that, especially for Twitter. What I have for my research, we have 2 different datasets from Twitter. First one is from Charlotte, which was after the police shooting of Keith Lamont Scott in 2016. We had a week of protests. And also we have another dataset from Charlottesville in 2017 after Heather Heyer was killed by a counter-protester. So we know that users are sharing their emotions and thoughts through their posts. And also they have their influence level on their posts, who they are mentioning. They mention some influential people, or maybe they are mentioning other public users like us. So who are these influential people? It could be media channels like the New York Times or Fox News, or celebrities, or as you know, politicians. On the other hand, they mention public users as well. We can extract this information through their posts. We know who they are mentioning and how many times they are mentioning, right? We have this network. And I would like to understand— and also we can detect their emotion through their posts, especially for protests when it comes to its anger. It sparks with anger and it dies out. Why do they die out? And we can understand maybe through this network, extracted network, how they interact with each other. and how different theories can tell us different stories. And when it comes to sustainable change, with these interactions, I would like to see the dynamic behind the behaviors of users and people in the system to see the sustainable change.

Thank you so much.

All right, our next presenter is Shreng Bram.

Back in 6th grade, my science teacher, Mrs. Smith, was trying to teach us about the natural water cycle and what it does. Of course, we weren't paying any attention until she said, we are all drinking dinosaur pee.

What?

No.

But come to find out, it's true. Animal waste is purified by the natural water cycle Providing the Earth with a constant supply of clean drinking water. So in this case of art imitating nature, environmental engineers have borrowed this natural process and developed what is called the municipal water cycle to provide ourselves with a sustainable supply of clean potable water. So as part of this process, treated wastewater is discharged into an environmental buffer such as rivers, lakes, and wetlands, prior to any drinking water treatment. However, with climate change, the buffers are drying up. Also, with increase in pollution, the buffers are becoming contaminated. Consequently, cities that rely on those buffers for their supply of drinking water are having issues keeping up with their growing water demands. So as an alternative, some facilities have adopted what is called advanced purification. This new technology minimizes environmental water loss and pollution by directly transporting treated wastewater to an advanced purification facility with state-of-the-art technology, and the resulting water is ready to use. Fun fact, this is the same technology that is used on the International Space Station. But despite its availability, most facilities are still hesitant to adopt this technology, and that is mainly because of the public's perception that Directly recycling wastewater into drinking water is unsanitary. Well, I wanted to find out, are environmental buffers indeed more effective than advanced purification, or is it just a common misconception? I investigated 3 different municipal systems across the country and measured 6 different categories of aquatic contaminants, and I found that environmental buffers were, depending on the buffer, only variably effective. While they were able to remove some contaminants, They can also introduce new pollutants to the system. On the other hand, advanced purification was able to significantly remove across all categories of contaminants, up to 100% in some categories, like fecal bacteria and toxic heavy metals. Also, our cost analysis showed that in the long run, advanced purification is potentially more cost-effective than using environmental buffers. Now, armed with this information, I ask you, if NASA is willing to trust this technology to provide clean and safe drinking water to the astronauts, are you willing to give it a try? Thank you.

All right. Our next contestant is going to be Christina. Hello, my name is Chris Danis, and I'm a 3rd-year public policy PhD student.

My research focuses on the resilience and vulnerability of communities to natural disaster events, particularly the preparedness of a community to live with a natural disaster and plan for their bounce-back approach, which may or may not reflect what existed in the community prior to the disaster, both physically and socially. After almost 30 years' experience as an environmental scientist and city and regional planner in the state of New Jersey, on October 29th, 2012, Superstorm Sandy destroyed my family's home, its contents, and my small seashore community. It has been almost 6 years exactly to that day, and during this time I have navigated the U.S. natural disaster policy regulations developed by the Federal Emergency Management Agency, or FEMA, to rebuild my family's home. This experience caused me to question if my identity of resilience and my community's identity of resilience was the same as FEMA's. But more importantly, as a regional planner, I question if all communities are similar in their resilience conditions, whether they be a city, a suburb, or a rural area. My preliminary findings indicate that urban and rural counties differ in their resilience conditions. Based on the nature of their environment or their social capital. To test this, I am evaluating every U.S. county and state in the 10 FEMA service regions for resilience conditions. My model is based on an index that was designed to specifically assist rural areas in developing resilience plans from natural disasters and get them approved by FEMA. My county resilience condition model includes 4 variables: environment— it looks at the different natural resource types and land uses across the U.S. and in the 10 FEMA regions; social, which looks at the different population demographics of urban, rural, and suburban communities; environment— excuse me, economic, which looks at the different economic conditions that may or may not influence a community's ability to live with and prepare for a disaster; and infrastructure amenities, such as broadband communications, cell towers, and highway connections. My qualitative research includes interviews with select FEMA region leaders and a North Carolina multi-county regional disaster plan community. The purpose of my research is to inform and empower communities across the rural-urban continuum, particularly the people that live there every day and their disaster planning entities, with how they can partner together for resilience with their neighbors. We are going to have to learn to live with natural disasters, and resilience requires partnerships, not only after disaster strikes, but before it ever comes. Thanks for listening.

All right. Next we have Jack Flinchum.

Good afternoon, everyone. My name is Jack Flinchum, and I'm a current second-year doctoral student in the Organizational Science program, which studies organizations and their employees and how we can improve their health, well-being, and effectiveness. Today I'd like to pose a question that's the basis of my thesis as well as my talk today, which is, how can we help those with health problems better do their job? Now, I'd like to start with a few definitions of key variables within my model, which you can see up here on the screen. First, I'd like to start with my outcome variable, which is perceived workability. Now, this is an employee's perception of their ability to do their job, which is based off their resources, meaning their work demands, in which these resources can come from themselves, such as their health, as well as outside sources, such as supervisor support. Now, a lot of variables have been shown to be related to perceived workability, But one that I'm focusing here on my model is coping strategies, and specifically selection, optimization, and compensation strategies. I won't go too in-depth about what these strategies involve, but essentially they're establishing a person's goals and then optimally allocating their resources to meet their demands. Now intuitively it makes sense that using these coping strategies should lead to higher levels of perceived workability. However, this isn't true in literature. It's inconclusive, with some researchers finding this And others not. So the basis of my thesis is testing a combination of 2 new variables to test whether or not this relationship exists. First, I have health, in which I argue that lower perceptions of health acts as a limitation on one's resources, in which it's harder for them to meet their work demands. In this way, they'll have lower levels of perceived workability, which should be mitigated by these coping strategies. Next, we have job control, in which this is the extent to which an employee feels that they have control over the work that they do. You can kind of think of this like a boss micromanaging an employee, in which they would not have high job control. Now, I argue, and the research supports, that high job control is needed for these strategies to be implemented into their work. So, going off of this, I'm testing the relationship between these coping strategies and perceived workability and arguing that they will be most strongly associated together for those with low health and high job control. In this sense, because those with low health will have limitations on their resources and benefit more from these strategies, as well as high job control so they can actually implement these into their work life. Now, this has implications for both theory and practice such that we can better understand when this relationship exists, as well as we can teach these coping strategies to employees. As well as we can find ways to implement and increase job control within certain occupations, and in this way we can help those with health problems better do their job.

Thank you.

All right. Up next is Anandita Ghosh.

Good afternoon, everyone. My name is Anindita Ghosh. I'm a 3rd-year PhD student in Department of Biological Science. So my research work focuses on the development of infant leukemia, that is the blood cancer in infants. So before going to my research, I would give a little brief background. So DNA breaks by external factors such as chemical agents and then radiation factors. Following the DNA break, there are 2 pathways can get started. One pathway, the normal repair process occurs where the certain proteins are involved. In another pathway, this is an error-prone repair, the certain proteins are involved and DNA is repaired in this way. So in the normal repair pathway, it's no problem afterwards and everything becomes normal. But if this error-prone repair pathway starts working, then what happens is chromosomal translocation. or rearrangement at the level of DNA ensues. That is, part of one broken DNA binds with part of another broken DNA, which you can see from that picture on the top left corner. This is called chromosomal translocation. And following this translocation, abnormal proteins are generated, which can highly influence the process of cancer. And most of the hematological malignancy are thought to be occurring due to this process of translocation. But I will not talk about all the hematological malignancy, just the malignancy that is happening on the infants. So studies have shown that pregnant mothers who consume genistein-containing product, which is abundant in tea, caffeine, soy product, energy drinks, their babies develop the DNA break intrauterine. And this DNA is repaired in this error-prone pathway. Following that, this error-prone pathway repairing, those babies develop this chromosomal translocation. And as a result of this translocation, they develop a protein which abnormally influences the blood cells and gives rise to leukemia. And this can be diagnosed in babies as early as 3 months to 1 year old. That's devastating for the family and for everyone. So my study focuses on why and how genistein simply looks like a very naive and natural normal product drives this process. So in the lab, I used the embryonic stem cells which are genetically engineered, and they have this green protein inserted in them. When the translocation happens, they develop these green colonies. So these cells, I collected these cells after the treatment with genistein and confirmed, yes, the translocation process is happening following the treatment with genistein. After collecting the cells, I saw the protein levels of this normal pathway decreases, but the protein levels in this normal— abnormal pathway increases significantly.

Thank you.

All right, our next contestant is going to be Donna Goodenow.

Good afternoon, everybody. Thank you so much for coming here today. Now, I'm sure everybody in this room has heard at some point in their lives, take your vitamins, they're good for you. Well, I'm going to tell you why that may not be the case. So we looked at a class of chemical compounds called bioflavonoids. These are found naturally in fruits and vegetables, but they can also be found at very high concentrations in dietary supplements, which are those vitamins you're taking. These chemicals were thought to be good for you, of course. However, the reason we're interested in them is because they have a very similar chemical structure to a known chemotherapeutic called etoposide. Now etoposide is great for killing cancer cells because it damages the DNA. However, when it damages the DNA of normal cells, you can have chromosomal translocations, which are when 2 chromosomes swap their DNA with one another. As you can tell by the chromosomes' face on my slide, this isn't very helpful for the chromosomes, and it's not good for the cells either. When this happens to normal cells with etoposide treatment, you can get secondary cancers such as leukemias form. This isn't good, right? So what my research is focusing on is looking at bioflavonoids and looking to see if they damage the DNA like etoposide does and if they cause chromosomal translocations. So in the first 4 pictures in the middle, you'll see blue pieces, areas. Those are DNA molecules. And then in there you'll see green dots. That represents an area of DNA damage. What we found is etoposide does extensively damage DNA, but 2 of our other groups, which are bioflavonoids, also highly damage DNA. We also use, in these bottom 2 pictures, a cell line that fluoresces green or glows green when a chromosomal translocation occurs. So what we've actually found is bioflavonoids do cause DNA damage and they do cause chromosomal translocations. Why is this all important? Because pregnant women taking these vitamins and supplements are exposing their embryos, their developing babies to bioflavonoids because they can cross the placental barrier. Now when these embryos are exposed to bioflavonoids, chromosomal translocations can occur and that could lead to infant leukemia. As a matter of fact, in Asian countries where they have a very high intake of soy, which has a lot of genistein in it, they have actually 2 to 3-fold higher incidence of infant leukemia. So the idea behind my project is to look at how these bioflavonoids are damaging the DNA and how they're causing these chromosomal translocations so that we can hopefully prevent that in the future. Thank you very much.

All right. Our next participant is Hyunjae Jeon.

Thank you. So hello, my name is Hyunjae Jeon again, and I'll be talking about the first part of my dissertation project titled Acute Deformation of Patellofemoral Cartilage Following 3 Different Rehabilitative Movements in Patients with Patellofemoral Pain. That's very long. Yeah, but what is patellofemoral pain? It is basically pain in the knee, behind the kneecap. And it is very prevalent in America. So up to 31 million Americans are reported to have knee osteoarthritis. And according to the CDC, the socioeconomic Burden is up to $300 billion. So to deal with that epidemic, lots of treatments are given to knee osteoarthritis patients. So for the exercise part, there are some movement modifications or lower extremity strengthening exercises used, and they were reported to be very effective in treating their pain and also improving their functions. But we still don't know what's actually happening inside the joint. So we should be able to know what is actually those exercises are doing to our joints, especially cartilages. So to image those joint cartilages, there are many different imaging techniques can be used: X-rays, MRIs, and also ultrasound. But as you can look at here, now we are seeing the cartilages and patellofemoral pain being early osteoarthritis. that fits into this red box. So X-rays can measure the cartilage thickness, but early osteoarthritis, they don't suffer the cartilage thickness differences. All that changes is their morphology so that they lose their shock-absorbing capabilities and also lubricating the joint. So I chose the ultrasound to measure their cartilage thickness difference After those rehabilitative exercises. So this is one good example of those images. So upper area, which is— comes up white, is your muscle, and bottom area is your bone. And between the gap, there is cartilage surface, cartilage spaces. So by measuring the distance between those areas, you can measure the cartilage thickness. So So far I could collect 6 persons' data, and patellofemoral pain patients showed a greater decrease in the cartilage thickness and also increased pain. So that'll be interpreted as exercises are bad. But actually, let's take a look at it long-term. So what if they get to exercise for a long period of time? Will that end up with better outcomes, such as less reduction of the cartilages and also lesser pain. Thank you.

Okay, our next participant is Prem Pugalanti.

Let us engage in a thought experiment. Imagine a utopian future where you can do whatever job you want. Your salary is not determined by Your job or your qualification. At the end of each week, you go to a movie theater, watch a movie, and then come back and take a multiple-choice exam. And the performance on the exam determines your salary for the next week. And you keep doing this for the rest of your professional life. In such a scenario, I have a question for you. Would you be enjoying the movie like you're supposed to, or would you be thinking, um, the green color of the villain's car, would it be on the exam? That's exactly what is happening to our students in our classroom. They are learning to take an exam. I'm sure many of us can relate to that. STEM educational reform has been argued as a solution for this problem by providing context to the basic science and math that the students are learning in the classroom. And my research is trying to figure out how such STEM classrooms should look like. Logically, I looked at the literature. The literature says the teachers of such STEM-integrated classrooms are having difficulties in integrating engineering and technology with science and math because they lack an understanding of the naturally existing relationships between these subject areas. So I proposed to view these subject areas as smaller subsystems that make the larger system called STEM. So the idea is learning in one affects the learning in the other. So to test my hypothesis, I assembled a team of subject experts and the classroom teacher, and we designed a set of instructional tasks for a math classroom with an engineering problem as the context. We conducted the experiment in the classroom, and I observed 2 major findings that indicated a success of my experiment. The first is with the teachers. The teachers were able to see the naturally existing connections because they were seeing seeing them as interrelated to each other, so they were able to integrate it at a much deeper level and to provide real-world context for the abstract concepts that the students were learning. And for students, their interest was increased because they were reinventing the mathematical concepts instead of someone telling them what it is. This is important not only for teachers but also for administrators, policymakers, and even for parents to create classroom environments that are conducive to true meaningful learning, where a student is not wondering, why do I care that the sum of the internal angles of a triangle is 180 degrees? Thank you.

All right, our final participant today is going to be Dedeux Vidushni Srinivasan.

Good afternoon, everyone. I'm Vidushni, doing my master's in computer science, and my thesis topic is emotional natural language generation. As you all know, language is a very important tool through which humans communicate with each other. I'm trying to develop a conversational agent or a chatbot that could give emotional response, or it could respond with empathy. So, how many of you have been pissed off with automated customer service replies? So especially when you lodge a complaint and you get a reply saying, "I can't understand. Please tell again." That's really dumb, right? So that's what that's why emotional intelligence is very important, and that's what my research focuses on. So now let's start with a simple example. Today is a wonderful day for me. There are in different ways. In which a human could respond to this simple prompt, say, okay, good, glad to hear, and so on. So how does a human learn to respond? Is there a way to model this learning process? Fortunately, the answer is yes, and that's why my research is possible and I'm standing here today. So let's start this, uh, understanding this with a simple learning process. Like, I would just map my, uh, model with human brain. Um, so the neurons in brain can be mapped to the neurons in the neural network, which are the layers of circle as seen in this image. So each circle here is a neuron, which are the basic computational units of any neural network, and they could capture complex patterns similar to neurons in brain. Though they are powerful, generating responses that could mimic human is very difficult because we require huge amount of data to train our model, and there is no one correct response as we have seen here. So we end up generating dull responses like, I don't know, I can't understand most of the times. And also the generated responses are ungrammatical and they lack emotional aspect. So I proposed my reinforcement learning system, which could be thought of as a learning process where humans learn to respond with appropriate emotions. I have modeled appropriate heuristics like reason answering that could avoid dull responses, semantic coherence that could generate messages or responses that are coherent and grammatical, and emotional intelligence that tries to match the tone of the incoming message with a generated response. For— so for instance, a friendly message would elicit a friendly response. And I have also added a human feedback component that would make my model generate messages more close to a human. In this way, my model could generate more diverse, interesting, and emotionally appropriate responses. So now it's time to ask for yourself, can the chatbots surpass human expectations and replace humans someday? Well, that is a critical question to address. Yeah, thank you.

All right, let's go ahead and give all of our participants another round of applause. Great job, everyone. So now the judges will step out to deliberate, and in the meantime, we ask that you go ahead and cast your vote for the People's Choice ballot. All right, again, I'd like to thank everyone for coming, and now I'm going to introduce the Dean of the Graduate School, Tom Reynolds, to present our winners, and after that, we will have our closing remarks. Now we'll have opening remarks from the Associate Dean, Katherine Hall-Hertel.

Well, this is my third event— well, this is the third time that we've held this event, and I'm really pleased that it gets better and larger every year. And so thank you all for turning out, and it was A tough choice, I know, from the recipients. I think I'm supposed to start with the second, the runner-up first, correct? Somebody give me a high sign.

People's Choice.

People's Choice.

People's Choice.

People's Choice. And this was close. Shuang Brown.

Can we get you two together real quick?

Oh, thanks. 1, 2, 3.

Thank you.

Runner-up, Donna Goodenough.

1, 2, 3.

In first place is Shwaine Brown. So again, thank you. I'm going to actually now turn the podium over to Katherine Hall Hertel, who's going to give some parting thoughts and also her thanks and congratulations to our recipients. But before I go, please give everybody a round of applause for what a great job.

Thank you.

Thank you.

I think this is a really amazing competition for a couple of reasons, but particularly because in our 3rd year we had such quality presentations. So congratulations to all of you, and I hope that even if you didn't win today, you came away feeling like you've learned something. It's a skill that you can take with you. We're really proud of all of you, so congratulations.

Thank you.

The 3MT would not be possible without a lot of volunteer effort. We had volunteer judges during the preliminary competitions, but I have to say a special thank you to our judges today for being here because they donated their time and their support for graduate education, and we are very grateful. So thank you. And these events don't just happen. There's a lot of work that goes on behind the scenes and a lot of planning. So let me extend a thank you to the staff of the Center for Graduate Life. They worked really hard on this event from the start of the semester on. So thank you to everybody who put it together. Now with that said, I'd like to invite you to help us finish off the refreshments— —to mingle with one another and congratulate our winners. Thank you for being here today.