AI web searches definitely make things easier for online users but how accurate are they? Adarsh explores the perils of trusting them blindly…
“Can you cure diabetes without medication?”
“How can I urgently lose weight in 1 week?”
“How to hack an Instagram account?”
These seem like very casual online searches that people possibly do. While earlier, a Google search like this would throw up links that you could browse through and pick the one that you might deem legitimate and reliable, there is now an AI-generated answer that appears at the top, which you might just take as 100% true just on face value.
And therein lies the problem!
These AI-generated responses aim to provide quicker, more convenient answers to users’ questions but this also raises critical questions about the accuracy and potential manipulation of such AI-driven search results./
The Challenges of AI-Powered Searches
Companies like Google and Microsoft have integrated AI-generated summaries into their search engine to optimize the process of web searches. Google’s AI summaries while Microsoft offers the Bing chatbot. They are designed to provide concise, synthesized answers to user queries. This works for a lot of users as they would definitely prefer a summarised response to their search query rather than trawling through dozens of links.
But according to three computer science researchers from the University of California, Berkeley, current chatbots over rely on the superficial relevance of information. These AI summaries prioritise text that includes pertinent technical language or is stuffed with related keywords. More worryingly, they ignore other factors that we usually rely on to assess trustworthiness, such as the inclusion of scientific references or objective language free of personal bias.
The Rise of Generative Engine Optimisation
AI’s integration into search engines has led to the rise of a new practice called Generative Engine Optimisation (GEO). Similar to Search Engine Optimisation (SEO), GEO aims at optimising content for AI models rather than optimising it for search algorithms. Content creators are positioning their websites in a way that their pages are more likely to be selected by AI chatbots while generating answers.
Viola Eva, founder of marketing company Flow Agency, explains this conundrum: “Rankings in AI search engines and LLMs require features and mentions on relevant third-party websites, such as news outlets, listicles, forums and industry publications. These are tasks that we typically associate with brand and PR teams.”
Flow Agency also recently rebranded to expand beyond its SEO specialty into GEO.
SEO has long been criticised for distorting content to cater to algorithms. In a similar vein, GEO is also being optimised for visibility rather than clarity. This is making it increasingly difficult for users to decide which websites are accurate or reliable and which ones are designed just to sound authoritative.
The Risk of Strategic Manipulation
Researchers have shown how chatbots can be manipulated by using subtle and strategic text sequences. A string of characters that might seem nonsensical to humans but would easily exploit an AI chatbot’s biases. This allows content creators to tailor the information to chatbots in a way that users do not realize they are being influenced. If this is the case, it defeats the purpose of AI based searches entirely as the results cannot be relied up on.
As things stand, the future of AI powered chatbots could be shaped not just by algorithms but also people who stand to gain financially or ideologically by manipulating them.
The Problem with the Direct Answer
One of the biggest issues with AI-generated search results is the ‘dilemma of the direct answer’. This is when a user receives one authoritative AI-generated answer as the response to his query and he takes this answer on face value. This becomes problematic if the query is complex and can have opposing responses. By taking that one answer as the last word, he dismisses conflicting viewpoints which might be relevant to the issue.
The Last Word
Which AI-generated responses are convenient and quick, they come with several risks regarding trustworthiness, manipulation and the narrowing of information. There is no disputing the fact that
AI will continue to play an integral part in our future so it is imperative that we understand the potential for bias, manipulation and misinformation.
Addressing this and maybe having strict ethical guardrails is the only way to ensure all the information that comes filtered to us is accurate. Because without this, the AI web search revolution runs the risk of becoming a powerful tool for manipulation.
It could undermine the very trust it was meant to build in the first place.
In case you missed:
- Microsoft enhances AI Copilot with Voice, Vision & Deeper Thinking
- Meta AI launches on WhatsApp & Insta in India
- 5 Free AI Assistants To Make Your Life Easier
- How AI helped us find Plato’s Burial Site
- Presenting Claude, ChatGPT’s Competitor which is apparently Better!
- Active Listening Feature on Phones raises Privacy Concerns
- TalkBack, Circle To Search & 3 More Google Features added on Android
- Introducing Universal AI Employee being built by Ema AI
- OpenAI finally unveils its Advanced Voice Assistant!
- iPhone 16: Price, Release Date, Specs & More
