Who wants to have to speak the language of search engines to find what you need?
It’s about a decade since I gave up on Google search. While I use Google services extensively for work and other areas of my life, search and personal email are not two of them. Instead, I use DuckDuckGo and, more recently, Perplexity Pro.
The latter is excellent, bypassing advertising and paid placements, acting as a natural language search agent for synthesising information. I tend to use it for information that would take several searches. Yesterday, for example, I gave it the following query: “I need a tool that can automatically take screenshot of a web page and then stitch them together. It should then make an animated gif, scrolling through the page from top to bottom. The website requires a login, so ideally it should be a Chrome browser extension." It gave me several options, approaching my request from multiple angles as there wasn’t a solution that did exactly what I needed.
Although this article in MIT Technology Review mentions Perplexity, it weirdly focuses mainly on Google and OpenAI. There’s no mention that you can choose between LLMs in Perplexity (I use Claude 3.5 Haiku) and the two issues it raises are copyright and hallucinations, rather than sustainability and privacy. Claude 3.5 Haiku is one of the lighter weight models when it comes to environmental impact, but it still consumes a lot more energy (and water, to cool the data centres) than a single DuckDuckGo search.
And then, when it comes to privacy, while it’s great that an LLM can personalise results based on what it already knows about you, there’s an amount of trust there that I’m increasingly wary of giving to companies like OpenAI. I cancelled and then resubscribed to ChatGPT last week. I’m not sure how long I can stomach the Sam Altman circus.
Ultimately, agentic search, where you ask a question in natural language and it shows you the sources it used to synthesise the answer, is the future. Perplexity seems pretty fair in this regard, pulling in my colleague Laura’s post as part of a response about the way that technology has shifted power over the last century. For me, this kind of thing is even more of a reason to work in the open.
There’s a critical digital literacies issue here, one that’s hinted in the last paragraph of the article (included below) and discussed in Helen Beetham’s podcast episode with Dan MacQuillan, author of Resisting AI: an Anti-fascist Approach to Artificial Intelligence. When “the answer” is presented to you, there’s less incentive to do your own work in finding your own interpretation. I think that is definitely a risk. Although, given that the internet is a giant justification machine already, I’m entirely sure it will necessarily make things worse — just perhaps make people a bit lazier.
The biggest change to the way search engines have delivered information to us since the 1990s is happening right now. No more keyword searching. No more sorting through links to click. Instead, we’re entering an era of conversational search. Which means instead of keywords, you use real questions, expressed in natural language. And instead of links, you’ll increasingly be met with answers, written by generative AI and based on live information from all across the internet, delivered the same way.
More to the point, you can attempt searches that were once pretty much impossible, and get the right answer. You don’t have to be able to articulate what, precisely, you are looking for. You can describe what the bird in your yard looks like, or what the issue seems to be with your refrigerator, or that weird noise your car is making, and get an almost human explanation put together from sources previously siloed across the internet. It’s amazing, and once you start searching that way, it’s addictive.
[…]
Sure, we will always want to use search engines to navigate the web and to discover new and interesting sources of information. But the links out are taking a back seat. The way AI can put together a well-reasoned answer to just about any kind of question, drawing on real-time data from across the web, just offers a better experience. That is especially true compared with what web search has become in recent years. If it’s not exactly broken (data shows more people are searching with Google more often than ever before), it’s at the very least increasingly cluttered and daunting to navigate.
Who wants to have to speak the language of search engines to find what you need? Who wants to navigate links when you can have straight answers? And maybe: Who wants to have to learn when you can just know?
Source: MIT Technology Review
Image: Kathryn Conrad