Facing the flip side of Hate on the frontlines of cyberspace | Hyeok Gyu Kwon | TEDxSNU

Facing the flip side of Hate on the frontlines of cyberspace | Hyeok Gyu Kwon | TEDxSNU

by Hyeok Gyu Kwon · View the channel on PSI

Applied Sciences & Engineering AI & Society UBDigital and Information technology: general topicsUBJDigital and information technologies: social and ethical aspects
🎙 Hyeok Gyu Kwon 👥 44.7M 📅 October 13, 2026 ⏱ 20 min 👁 3 📄 expert opinion 🧭 2026-10-13
Available in: English (current) Français

Keywords

hate speechAI monitoringRPAnegativity biasonline disinhibition

Summary

The talk presents the story of Hyeok Gyu Kwon, CEO of Mero Co., Ltd., who developed an AI-based service to detect and respond to online hate comments. He begins by recalling his own frustration when a favorite YouTuber faced an onslaught of malicious comments, leading him to write a simple script to collect them for legal action. This initial effort evolved into a business that combines crawling, RPA, and AI to automate the detection, evidence collection, and temporary removal of hateful content. Kwon explains the technical process: selecting keywords, choosing target websites, and defining the scope of collection. He emphasizes that AI helps tailor classification to each client’s specific criteria, while humans make final decisions. Drawing on data collected, he challenges the common belief that most internet comments are hateful, attributing this misconception to negativity bias. He profiles typical hate commenters as ordinary people who act differently online due to reduced feedback and social cues. Kwon discusses how platforms’ reward systems (likes, views, monetization) amplify hostile behavior and how algorithms create filter bubbles, reinforcing extreme views. He concludes that the key is to pause before judging and to seek understanding, using a personal anecdote about his girlfriend’s ‘말랑이’ (squishy toy) to illustrate how curiosity can replace prejudice. He urges the audience to question whether their immediate perception is the whole truth, as love and understanding require effort and time.

245 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides insight into a real-world application of AI for social good, offering a hands-on perspective on how technology can be harnessed to combat online hatred. The speaker’s argument is logically structured: from problem identification, to technical solution, to psychological and social analysis. He effectively uses personal examples and simple analogies to support his points, though the evidence is largely anecdotal. The discussion of negativity bias and filter bubbles is credible but lacks scholarly citation. The reasoning that ordinary people become hateful online due to environmental factors is persuasive but not deeply theorized. Overall, the argumentation is coherent and compelling, but it would benefit from stronger empirical or academic backing.

Scientific Rigor, Source Quality, Title Accuracy

The speaker is the founder of a company, so his claims are based on practical experience, but this also introduces potential bias. No external sources are cited in the talk; the only reference is the standard TEDx link to explain the event format. The title aligns well with the content, emphasizing both the ‘frontline’ (the practical battle against hate) and the ‘flip side’ (the nuanced understanding of both perpetrators and victims). The technical explanations are clear but not overly detailed, and the psychological concepts are mentioned without rigorous citation. The talk’s strength lies in its authenticity and personal touch, while its weakness is the absence of vetted references.

234 words

Title / Content Match

The title accurately reflects the talk's focus on addressing online hate from a frontline perspective, with an emphasis on understanding the human and technical dimensions of the issue.

Quality & Reliability

6/10

The talk is an expert opinion based on the speaker's entrepreneurial experience in building an AI-powered anti-hate service. It lacks peer-reviewed references and relies on personal anecdotes, limiting its scientific rigor.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • TEDx program information — The talk was given at a TEDx event, and this link explains the TEDx format and community organization.

Contribution & Novelties

The talk offers a unique practitioner’s view on using AI for anti-hate efforts, combining technical workflow with psychological insights. The novelty lies in the practical integration of crawling, RPA, and AI classification for legal evidence, and the counterintuitive finding that non-hate comments dominate. This contrasts with common media narratives.

Pour aller plus loin :

  • Negativity bias — Relevant to the speaker’s explanation of why negative content dominates perception.
  • Filter bubble — Central to the discussion of algorithmic amplification of hate.
  • Online disinhibition effect — Explains why ordinary people may engage in hostile behavior online.
  • Confirmation bias — Underpins the idea of echo chambers and polarization in online spaces.

108 words

Radar Profile

The radar profile indicates a moderate performance across all dimensions, with slightly higher scores in quantity of information but lower in technical depth and reliability. This suggests a talk that is informative and accessible but not highly rigorous or technically advanced, fitting a personal-experience TEDx format.

Reliability 5/10