Populating Commonsense Knowledge Bases (CSKB) is an important yet hard task in NLP, as it tackles knowledge from external sources with unseen events and entities. Fang et al. (2021a) proposed a CSKB Population benchmark with an evaluation set CKBP v1. However, CKBP v1 adopts crowdsourced annotations that suffer from a substantial fraction of incorrect answers, and the evaluation set is not well-aligned with the external knowledge source as a result of random sampling. In this paper, we introduce CKBP v2, a new high-quality CSKB Population benchmark, which addresses the two mentioned problems by using experts instead of crowd-sourced annotation and by adding diversified adversarial samples to make the evaluation set more representative. We conduct extensive experiments comparing state-of-the-art methods for CSKB Population on the new evaluation set for future research comparisons. Empirical results show that the population task is still challenging, even for large language models (LLM) such as ChatGPT. Codes and data are available at https://github.com/HKUST-KnowComp/CSKB-Population.

Harmful content is pervasive on social media, poisoning online communities and negatively impacting participation. A common approach to address this issue is to develop detection models that rely on human annotations. However, the tasks required to build such models expose annotators to harmful and offensive content and may require significant time and cost to complete. Generative AI models have the potential to understand and detect harmful content. To investigate this potential, we used ChatGPT and compared its performance with MTurker annotations for three frequently discussed concepts related to harmful content: Hateful, Offensive, and Toxic (HOT). We designed five prompts to interact with ChatGPT and conducted four experiments eliciting HOT classifications. Our results show that ChatGPT can achieve an accuracy of approximately 80% when compared to MTurker annotations. Specifically, the model displays a more consistent classification for non-HOT comments than HOT comments compared to human annotations. Our findings also suggest that ChatGPT classifications align with provided HOT definitions, but ChatGPT classifies "hateful" and "offensive" as subsets of "toxic." Moreover, the choice of prompts used to interact with ChatGPT impacts its performance. Based on these in-sights, our study provides several meaningful implications for employing ChatGPT to detect HOT content, particularly regarding the reliability and consistency of its performance, its understand-ing and reasoning of the HOT concept, and the impact of prompts on its performance. Overall, our study provides guidance about the potential of using generative AI models to moderate large volumes of user-generated content on social media.

AI developments from generating videos, voices, pictures and human-like conversations are growing rapidly. Lawmakers say they are trying to keep up. Reddit is planning to charge tech companies for the use of its application programming interface. Through the method, outside entities have the ability to download and process the social network's conversations, including tech giants using them in the development of artificial intelligence tech like large language models. "The Reddit corpus of data is really valuable," Reddit co-founder and CEO Steve Huffman told The New York Times Tuesday.

Check out what's clicking on Foxnews.com. Leaders of the U.S. Federal Trade Commission said on Tuesday the agency would pursue companies who misuse artificial intelligence to violate laws against discrimination or be deceptive. The sudden popularity of Microsoft-backed OpenAI's ChatGPT this year has prompted calls for regulation amid concerns around the world about the possible use of the innovation for wrongdoing even as companies are seeking ways to use it to enhance efficiency. In a congressional hearing, FTC Chair Lina Khan and Commissioners Rebecca Slaughter and Alvaro Bedoya were asked about concerns that recent innovation in artificial intelligence, which can be used to produce high quality deep fakes, could be used to make more effective scams or otherwise violate laws. FTC Chair Lina Khan testifies on Capitol Hill in Washington on April 21, 2021.

On Sunday, the New York Times reported that Samsung, which makes more smartphones than any other company, has considered switching its devices' default search engine from Google to Bing, thanks in part to the excitement around Bing's AI features. The Times reported that threat sparked "panic" at Google, whose name is synonymous with online search.

Snapchat's ChatGPT-powered AI personality is expanding to all the app's users. An upgraded version of "My AI," the in-app chatbot that was previously limited to Snapchat subscribers, is now launching globally, Snap CEO Evan Spiegel announced at the company's Partner Summit event. With the expansion, My AI has a number of new Snapchat-specific features. It can provide Snapchat users with recommendations for restaurants and other activities based on what's popular in the Snap Map, and can suggest augmented reality lenses. Users can also add the AI to group chats, and set a custom name and avatar (via Bitmoji) for the AI persona.

These images provide clues as to how ChatGPT will look if its creators make a physical version of their hugely popular artificial intelligence. The maker of ChatGPT, OpenAI, has invested in 1X, a company that makes humanoid robots designed to do human jobs after the success of the online chatbot. The robot, named EVE, has manipulators which can pick up objects and pack and unpack boxes - and is designed to work alongside human beings. OpenAI's Startup Fund led an investment round that raised $23.5 million for the 1X robot, which is set to hit the market this summer. The investment fuels OpenAI's rivalry with Elon Musk's Tesla bot, which has yet to begin production.

Figuring out how humans and robots can collaborate to effectively carry out tasks together is a rapidly growing area of interest. For successful collaboration between humans and robots, communication is key.

OpenAI has just over a week to comply with European data protection laws following a temporary ban in Italy, and a slew of investigations in other EU countries. If it fails, it could face hefty fines, be forced to delete data, or even be banned. But experts have told MIT Technology Review that it will be next to impossible for OpenAI to comply with the rules. That's because of the way data used to train its AI models has been collected: by hoovering up content off the internet. Since the pandemic closed schools in 2020, nearly all students have been learning on school-issued laptops or tablets. But many experts suspect that the technology may be changing how they read, as reading on a screen is fundamentally different from reading on the page.

The AI landscape has started to move very, very fast: consumer-facing tools such as Midjourney and ChatGPT are now able to produce incredible image and text results in seconds based on natural language prompts, and we're seeing them get deployed everywhere from web search to children's books. However, these AI applications are being turned to more nefarious uses, including spreading malware. Take the traditional scam email, for example: It's usually littered with obvious mistakes in its grammar and spelling--mistakes that the latest group of AI models don't make, as noted in a recent advisory report from Europol. Think about it: A lot of phishing attacks and other security threats rely on social engineering, duping users into revealing passwords, financial information, or other sensitive data. The persuasive, authentic-sounding text required for these scams can now be pumped out quite easily, with no human effort required, and endlessly tweaked and refined for specific audiences.