Home » Uncategorized

Search Engines vs Synthesis Engines: The Future of Search

  • ajitjaokar 
reading-297450_1280

With the announcements last week from Microsoft and openAI, we are now all actively discussing the future of search

Here are some key takeaways as I interpret them:

  1. Search engines could now be ‘answer engines’: The idea is not new, but with LLMs, the quality of answers and suggestions gets better
  2. Chat could be an integral part of search: search becomes interactive and iterative ie unlike conventional search engines, the first response is now the starting point instead of being the ending point
  3. A co-pilot for the web: The Edge browser becomes a copilot for the web i.e. it can provide features like summarization for a website. 
  4. LLM driven search is embedded into other products. 
  5. Prompt engineering is a new skill for training and accessing search: Through Prompt engineering, a series of tasks can be converted to a  prompt-based dataset that a LLM can learn and respond to.

More interestingly, Balaji Srinivasan shared an interesting idea: search engines could evolve into synthesis engines. Through prompt engineering, you can provide a sequence that composes a complex response. In contrast, search engines retrieve a response to single-word queries. So, the future lies in composability. Already, many complex tasks are composable – such as writing scripts and building simple web apps. 

So, a legacy search engine will be used to cite someone but you will use a synthesis engine for your own creations.

Most people agree that this future is not far. Whatever your viewpoint on the timelines, I think we all agree that innovation in search is long overdue, and ultimately, innovation will benefit consumers. 

We live in interesting times! 

source: tweet from Balaji Srinivasan

image source: https://pixabay.com/vectors/reading-studying-lawyer-pastor-297450/