Generalized Zero-Shot Learning (GZSL) and Open-Set Recognition (OSR) are two mainstream settings that greatly extend conventional visual object recognition. However, the limitations of their problem settings are not negligible. The novel categories in GZSL require pre-defined semantic labels, making the problem setting less realistic; the oversimplified unknown class in OSR fails to explore the innate fine-grained and mixed structures of novel categories. In light of this, we are motivated to consider a new problem setting named Zero-Knowledge Zero-Shot Learning (ZK-ZSL) that assumes no prior knowledge of novel classes and aims to classify seen and unseen samples and recover semantic attributes of the fine-grained novel categories for further interpretation. To achieve this, we propose a novel framework that recovers the clustering structures of both seen and unseen categories where the seen class structures are guided by source labels. In addition, a structural alignment loss is designed to aid the semantic learning of unseen categories with their recovered structures. Experimental results demonstrate our method's superior performance in classification and semantic recovery on four benchmark datasets.

A characteristic feature of human semantic cognition is its ability to not only store and retrieve the properties of concepts observed through experience, but to also facilitate the inheritance of properties (can breathe) from superordinate concepts (animal) to their subordinates (dog) -- i.e. demonstrate property inheritance. In this paper, we present COMPS, a collection of minimal pair sentences that jointly tests pre-trained language models (PLMs) on their ability to attribute properties to concepts and their ability to demonstrate property inheritance behavior. Analyses of 22 different PLMs on COMPS reveal that they can easily distinguish between concepts on the basis of a property when they are trivially different, but find it relatively difficult when concepts are related on the basis of nuanced knowledge representations. Furthermore, we find that PLMs can demonstrate behavior consistent with property inheritance to a great extent, but fail in the presence of distracting information, which decreases the performance of many models, sometimes even below chance. This lack of robustness in demonstrating simple reasoning raises important questions about PLMs' capacity to make correct inferences even when they appear to possess the prerequisite knowledge.

A tweet pinned to the top of Hegde's feed in honor of Modi's birthday calls him "the leader who brought back India's lost glory." On January 7, the account tweeted a screenshot from ChatGPT to its more than 185,000 followers; the tweet appeared to show the AI-powered chatbot making a joke about the Hindu deity Krishna. ChatGPT uses large language models to provide detailed answers to text prompts, responding to questions about everything from legal problems to song lyrics. But on questions of faith, it's mostly trained to be circumspect, responding "I'm sorry, but I'm not programmed to make jokes about any religion or deity," when prompted to quip about Jesus Christ or Mohammed. That limitation appears not to include Hindu religious figures.

Large language models like OpenAI's GPT-3 are massive neural networks that can generate human-like text, from poetry to programming code. Trained using troves of internet data, these machine-learning models take a small bit of input text and then predict the text that is likely to come next. But that's not all these models can do. Researchers are exploring a curious phenomenon known as in-context learning, in which a large language model learns to accomplish a task after seeing only a few examples--despite the fact that it wasn't trained for that task. For instance, someone could feed the model several example sentences and their sentiments (positive or negative), then prompt it with a new sentence, and the model can give the correct sentiment.

Artificial Intelligence (A.I.) has garnered widespread attention from users and investors after the viral rollout of the AI-powered chatbot ChatGPT. To give you an idea of how successful the launch has been, ChatGPT is now the fastest consumer app to reach 100 million active users – taking just two months to reach the milestone. As a result of the meteoric rise to popularity, privately held ChatGPT creator OpenAI has secured more than $10 billion in investments from Microsoft. Companies like Amazonhave been using AI under the hood for years. For example, Amazon leverages AI on its back end to increase sales on its e-commerce platform (if you add a table to your shopping cart, it will suggest chairs).

Senior Microsoft executives on Tuesday unveiled plans to use AI capabilities to improve its struggling online search engine Bing, and its internet browser Edge. It's hoping to offer more competition to market leader Google's Search function and Chrome web browser. The announcement comes as the new artificial intelligence writing program ChatGPT enjoys widespread public attention following its launch last November. Microsoft had been a partner and 9% stakeholder of the OpenAI non-profit that created ChatGPT since 2019, but in late January it made another major investment in the group -- reportedly as much as $10 billion (roughly €9.3 billion) -- to increase that presence. Its redoubled interest in OpenAI is thought to be a bid to counter some of the wider research operations of Google's Alphabet Inc. parent company.

Microsoft Corp is revamping its Bing search engine and Edge web browser with artificial intelligence, the company said on Tuesday, in one of its biggest efforts yet to lead a new wave of technology and reshape how people gather information. Microsoft is staking its future on AI through billions of dollars of investment as it directly challenges Alphabet Inc's Google. That could mean new competition for business customers using cloud and collaboration products as well as a vigorous return to consumer markets where Google now leads. Working with the startup OpenAI, Microsoft is aiming to leapfrog its rival and potentially claim vast returns from tools that speed up all manner of content creation, automating tasks, if not jobs themselves. "This technology is going to reshape pretty much every software category," Microsoft Chief Executive Satya Nadella told reporters in a briefing at Microsoft headquarters in Redmond, in Washington state.

Since the hit sitcom Seinfeld went off the air in 1998 after nine seasons, the show's devoted followers have long mused about an alternate reality: What if the original "show about nothing" had never ended? Now they've gotten what they wished for--well, sort of. In mid-December, a never-ending AI-generated reboot, aptly named Nothing, Forever, launched on the streaming platform Twitch. You could, anyway, until earlier this week, when forever abruptly ended--or was at least briefly interrupted, and in just about the most fitting way imaginable: by the AI scriptwriter devolving into bigotry. Nothing, Forever is powered by Davinci, the newest publicly available version of OpenAI's GPT-3 language model--a close relative of ChatGPT--and although that technology is impressive, the show, in most respects, is not.

It has since announced a new premium tier of Microsoft Teams - its messaging software - that will feature ChatGPT, including a feature that automatically generates notes and highlights of meetings.

The age of generative AI is upon us, and this week alone Google and Microsoft made major announcements around their respective products for the masses. While Google unveiled an "experimental conversational AI service" called Bard yesterday, Microsoft had a fuller slate of news to share at its event in Redmond, WA today. Through a partnership with ChatGPT maker OpenAI, Microsoft is adding more advanced AI conversation models to power updates to Bing and Edge. The company's keynote today happened at breakneck pace, with demos whizzing by so quickly there was barely enough time to make sense of the updates. Thankfully, I was able to briefly check out a full demo here with Dena Saunders from Bing Engineering.