Hmm...kinda poor summary from ChatGPT 3.5...

• 🎮 A superhuman AI named Gobot won against the world champion in the board game Go in 2016.
• 🧑‍🤝‍🧑 In January 2023, AI researchers allowed an amateur human to beat one of these superhuman Go AI bots named Catego, with a nearly 100% win rate.
• 🤖 There are three main categories of artificial intelligence: Narrow AI, General AI or AGI, and superintelligence.
• 🤔 AI research today is focused on making AI better than narrow but not quite general intelligence, and far from sci-fi level AI.
• 🌐 Large language models, like Chat GPT, have been the subject of news stories almost every day, overshadowing the recent news that a human finally bested the best AI Go bot named Catego.
• 🤯 The flaw that allowed a human to beat Catego applies to all of today's most widely used AI systems, especially Chat GPT.
• 🔍 The fundamental problem is that we don't actually know how these systems work on the inside because that's how we design them.

The web page discusses the Harper.ai plugin, which allows users to use their chat jibiri to browse and extract data from documents, summarize web pages and YouTube videos, generate outlines, create comparison tables, and extract data in table format. Users can also select different models to use, depending on their needs. The plugin has a wide range of applications, and the possibilities are endless.

The CEO of OpenAI, Sam Altman, has confirmed that GPT-5 is not in training and will not be for some time. This is in response to an open letter calling for a pause on the training of AI systems more powerful than GPT-4. Altman emphasizes the importance of safety issues and caution in releasing GPT-5, as emergent abilities can arise spontaneously from the interactions and complexities within a system. OpenAI is testing and making GPT-4 even safer, building it on years of alignment research that they pursued in anticipation of models like GPT-4. Altman also agrees that OpenAI should release its alignment data set and alignment evaluations to other companies to ensure AI safety.

The article discusses prompt engineering, which is the process of creating inputs that determine the output generated by an AI language model. High-quality prompts can result in better output while poorly defined prompts can lead to inaccurate or negative responses. The article explains that AI language models rely on deep learning algorithms and natural language processing (NLP) to fully understand human language. There are two main learning methods for language models: supervised and unsupervised learning. The article also covers prompt categories, which include information-seeking prompts and instruction-based prompts