When engaged in ideation and problem-solving it can be difficult to differentiate between AI Tools, AI agents or heuristics. Simply, AI tools, such as ChatGPT work in a reactive manner in response to prompts whereas AI agents operates independently without human intervention. In contrast, heuristics are adaptive tools involving experimentation and trial and error. Or, to use an aviation metaphor, AI tools are like co-piloting, AI agents similar to auto-piloting, and heuristic tools comparable to flying manually. Key considerations when deciding between these tools include: task predictability; complexity of decision-making; and need for adaptability. The choice, however, depends on user needs. AI agents, for example, in automating decision-making processes, don't need step-by-step instructions but fall short of true autonomy. That is, in real-world applications, AI agents are faced with technical challenges and issues of trust and security. For everyday design thinking, this suggests using either iterative prompting to unlock the power of GenAI tools (a process of refinement) or using heuristic methods grounded in pragmatism and domain expertise, particularly to ensure decisions are made with ethical considerations and human oversight. However, designers typically use hybrid problem solving strategies and so prompt engineering may require heuristics too. Sources: https://relevanceai.com/relevance-academy/when-to-use-agents-v-tools https://www.ibm.com/think/topics/ai-agents https://mikecarruego.medium.com/choosing-the-right-algorithm-machine-learning-vs-heuristics-dc0b65e97d98
No comments:
Post a Comment