There is enough evidence to suggest that the average human attention span has been declining, with the most famous data point being 8 seconds. Are you still reading this, or did I lose you?
We are on a constant hunt for short content, quick summaries, and headlines. In contrast, the AI tools we are developing are learning to handle larger inputs (context window) and focus on different parts of the input data (self-attention). As we increasingly use AI, what does this mean for us?
The AI Side
Let's consider a scenario where we're deploying an AI agent for a generic business task. In the process, we would expect AI to:
- Read relevant data and use the takeaways as input for subsequent tasks
- Make some assumptions which may not be explicit to us
- Compress research time to levels we could not dream of
- Eventually, create outputs that are ready to go without extensive edits
The Human Side
While AI agents handle all of this, will we be able to deliver on our role?
- Spend more time understanding and constructing the problem
- Take the effort to codify the context for AI to make better assumptions
- Do a critical review of the work AI has delivered to ensure accuracy
- Review the final output to ensure the entire story comes through
AI will get better at its jobs with increasing context windows and attention, but will humans be able to keep their side of the attention bargain? The title was inspired by the research paper "Attention Is All You Need" published by researchers from Google in 2017.