Major tech companies like OpenAI and Anthropic face a problem: they want to operate their powerful AI models like Claude or Codex without costs becoming too high. A new solution is to make AIs express themselves very briefly. This means models should answer as simply as possible to save computing power. It's like teaching a clever speaker to only use bullet points.
This development shows that AI developers must pay close attention to costs. Every request to a large language model costs money. These costs can quickly reach millions. If companies get their AIs to 'speak' more sparingly, they save a lot of money. For you as a user or company, this means: The AI may no longer give a detailed answer. But it is much cheaper and faster to operate. This changes what is expected from AI models: Is good quality or fast work more important?
Companies like OpenAI and Anthropic are looking for new ways to reduce the costs of their large language models. The strategy is called 'Caveman'. In this approach, AI models like Claude or Codex are specifically trained to give very short and simple answers. A key OpenAI employee worked on this project to make the models more efficient. The idea is: every extra letter or word in an AI answer costs computing power. These costs add up with millions of requests.
For individuals, freelancers, and creators, this means: AI helpers like ChatGPT or similar models will be faster and cheaper. You might get your answers more directly and without many words. This saves time in everyday life or with creative tasks. At the same time, you have to get used to the AI no longer explaining things in such a 'human' or detailed way. Your workflow will become more practical. You get the facts, but less of the friendly chat.
Companies want to use AI on a large scale. Here, computing costs are very high. With the 'Caveman' approach, companies can make their AI projects much cheaper. This applies to all areas, from internal communication to customer service. Fewer 'superfluous' words in AI answers mean big savings. These directly impact profits. For competitiveness, this is important to make AI applications profitable in the first place. It's about offering good quality while still having affordable prices.
This strategy offers new opportunities for everyone who uses AI models. First, access to powerful AI becomes cheaper. This makes it easier for small companies and freelancers to get started. Second, the speed of responses can increase. This is particularly important for applications that need to react immediately. Third, developers must formulate their instructions, known as prompts, more precisely. This way, they get the desired information without unnecessary 'filler material'. This promotes more efficient communication with the AI. It saves time and money.
