Upgrade your prompting skills with a multishot approach and a conversational style with chatbots while keeping a healthy distance
Did you know that Claude and Alexa are the hottest couple at the moment?
While Apple is partnering with OpenAI to enhance Siri with ChatGPT, Amazon’s new Alexa will incorporate Claude AI. “At its core, talking to a model is very similar to talking to a person. Clear instructions and providing all relevant information are crucial to achieving the desired results,” says AI expert Zack Witten.
Anthropic, the company behind the Claude chatbot, has a strong ethical commitment to training AI and building robust, reliable neural networks. The engineers working for the company have shared top tips for better AI prompting, emphasising clarity and specificity. Think of prompting as setting GPS coordinates; providing examples can also enhance the accuracy of models.
Try multishot prompting to achieve better text-to-text results; this approach is also effective for other generative AI tools, such as text-to-image, sound, or video, depending on the platform. Include up to five examples to show the AI what kind of outcome you expect. Experts say that more examples can lead to better performance, so don’t be stingy.
If you want to practice, here are some multishot prompting examples for Claude.
Video AI Content is King (for Better or Worse)
The video world has been shaken by Sora, leading to numerous text-to-video generative tools popping up like mushrooms.
One such promising tool is Video-01 AI, the new video generator developed by MiniMax and backed by tech giants Alibaba and Tencent. This tool is capable of producing high-resolution videos with visually stunning nuances, as well as highly detailed and lifelike scenes. Of course, users will still be responsible for spotting hallucinations or correcting inaccuracies. No one expects we’ll ever really eliminate them, not even Apple’s CEO, Tim Cook.
Platforms are developing tools to protect creators’ originality as AI video content becomes more prevalent. “As AI evolves, we believe it should enhance human creativity, not replace it. We’re committed to working with our partners to ensure future advancements amplify their voices,” YouTube stated in a recent announcement.
YouTube is also very committed to taking down unauthorised AI-generated audio clones after over 200 artists, including Billie Eilish and Katy Perry, described it as an “assault on human creativity.”
Being Too Friendly with AI Can Get You in Trouble, but Good Manners Are Still Important
Next-generation voice assistants powered by Artificial Intelligence are becoming increasingly similar to Samantha from the film “Her,” capable of connecting with humans on a deeper and more personal level.
The new Empathic Voice Interface (EVI) launched by Hume AI demonstrates emotional intelligence and has a seductive tone of voice. The company is strongly committed to defining ethical guidelines for a more empathic and safe approach to AI, working alongside ethicists and social scientists on this issue.
Numerous studies on human-machine interaction show that relationships between humans and non-humans, such as chatbots, can lead to emotional dependence. Communicating with AI in natural language can create an inappropriate bond and a false sense of familiarity and closeness with a machine.
You can continue greeting ChatGPT when starting a conversation or thanking Siri or Alexa for their services while still remembering that they have no feelings or self-awareness. Kindness is what keeps us human, and it’s always the way to go.
Gaia Giordani
Generative AI explorer and New Media Communication expert.