· published on LinkedIn · translated from French

Answer engine: the new game-changer?

Web search is shifting from the search engine to the answer engine. What that implies.

In case you missed the news, the web search industry is in the middle of a shift. Roughly, we’re going from:

  1. I have a question,
  2. I look for resources that answer my question,
  3. I find my answer in those resources,

to:

  1. I have a question,
  2. A LLM searches and summarizes the information to answer the question precisely.

This is what we call an “answer engine”. ⚙️ OpenAI just launched SearchGPT, and Google has been working on similar solutions since the crisis mode triggered by the launch of ChatGPT in 2022.

Given that web search made Google what it is today, it’s no surprise that everyone is jumping in. But this change implies much more than simple automatic answers.

1. Toward omniscience? With this ability to answer a vast amount of questions, we’re seeing the emergence of systems that will end up resembling those described in sci-fi movies like “Transcendence” or “I, Robot”. Today, it’s mostly “reading” info and restituting/synthesizing it, tomorrow it will be real actions. Ex: today checking available trains, tomorrow buying the cheapest tickets. Guess: if you push this to the extreme, it could even lead to the rise of AI-first players and a gradual disappearance of web interfaces (yes yes, not for right now, no panic 😄).

2. Accelerating productivity and learning. Progress in a field is often driven by various factors: funding, human resources, motivation… Here, I want to talk about another factor: the speed of access to the answers we ask ourselves about the field.

Stack Overflow’s latest survey reveals that 81% of devs think AI boosts their productivity, and 62% believe it accelerates their learning. Devs are the first hit by these tools, and it will quickly reach many other professions.

We’re approaching a point where getting instant answers to our ultra precise questions becomes the norm. After language, the printing press and the Internet, we’re reaching a stage where access to information becomes nearly instant.

But here’s the thing, the question that remains is crucial: in this flow of generated answers, will we manage to make these systems reliable fast enough so they don’t lead us astray?