Recently we hear a lot about ChatGPT. It's the first time we've talked about it and therefore a short but complete introduction is needed. In the course of the article we will first of all go to see what the new OpenAI project is and what it is for. In addition to this, we will expand the discussion by introducing his "rival" Anthropic Claude. Curious that the latter was created by former employees of OpenAI.
Anthropic Claude's AI system modifies OpenAI's ChatGPT to align it with human intentions. How will he do?
What is ChatGPT
ChatGPT (GPT stands for Generative Pre-Trained Transformer) is an AI-powered chatbot system similar to automated customer support chats. However, the bot is much more advanced in accomplishing its tasks, as it's not limited to a handful of response options. These, we remember, sometimes close in a circle leaving us dry-mouthed.
Read also: LG launches an ultra slim and also completely wireless Smart TV
ChatGPT is much more complex. The working format allows you to ask questions while he (it feels weird to use a pronoun for an AI right?) admits his mistakes and refuses inappropriate requests. ChatGPT can do more than just have a conversation. OpenAI has equipped the chatbot with the ability to correct grammar, condense difficult text into simpler concepts, convert movie titles into emojis, and even debug Python code.
The chatbot is based on the OpenAI GPT-3 language model. This language model uses deep learning technology and is trained with algorithms that are fed large amounts of text from the internet. Users have described the technology as an alternative to Google for its ability to provide descriptions and answers to complex questions.
An introduction to Anthropic Claude
Anthropic, a company founded by former OpenAI employees created an AI similar to OpenAI's ChatGPT called Claude. The system is currently only accessible via a Slack integration as part of a closed beta test, and has reportedly been shown to have improvements over its aforementioned counterpart. To create Claude the company developed a technique they call “Constitutional AI”, which is an AI system that aims to align the system with human intentions by providing a principles-based approach, allowing the AI system to respond to questions based on a basic set of guiding principles similar to OpenAI's ChatGPT.
Read also: The WiFi signal can detect respiratory pathologies with very high accuracy
The "protocol" on which Claude is created begins with the creation of a list of about 10 principles that serve as a sort of "constitution" for the artificial intelligence system. The specific principles used in Claude's creation have not been publicly disclosed, but Anthropic says they are based on key concepts such as beneficence (maximizing positive impact), non-maleficence (avoiding harmful advice), and autonomy (respecting freedom of choice ). This approach aims to align the AI system with human intentions, allowing it to answer questions using a set of guiding principles.
Basically, Claude is a tool similar to other language models, such as ChatGPT, which is trained on a large amount of text data from the Internet to predict pattern-based words in the semantic context of the surrounding text. As a result, he can hold conversations, tell jokes and provide information on various topics.
What is the difference between ChatGPT and Anthropic Claude?
Yann Dubois, a PhD candidate from Stanford's AI Lab, also compared Claude and ChatGPT. He has found that Claude "generally follows more closely what is asked" but is "less concise" because he tends to explain his own answers and asks how he can help further. Claude scored better on some trivia questions, especially those related to entertainment, geography, history, and basic algebra, without providing the extra information that ChatGPT sometimes includes. Also, unlike ChatGPT, Claude admits (sometimes) when he doesn't know the answer to a question.
Through | TechCrunch
