Will AI End Search as We Know It?
Whether you like artificial intelligence (AI) or not, it’s here to stay. In small doses, it’s a fantastic educational and research supplement, but when gone rogue—and in the wrong hands—it’s a little nerve-wracking.
Given the astonishing recent strides in AI technology, it’s easy to understand why AI experts are calling for a time-out. As more powerful AI systems are developed, the potential risks to society are easy to spot.
We’ve talked extensively about the use of AI in marketing (more than once), but answer engines powered by AI have accelerated the direct, intentional use of AI in everyday life. What makes AI, well, AI is that it learns and gets smarter over time. AI is set up in neural networks, much like the human brain, and it gains competency with each progressive use. This is what makes it so powerful, and it has begun to complicate the marketing landscape—especially regarding search.
Take the most prominent example today: Open AI’s GPT4, one of the most powerful AI systems ever created. The big differentiator for GPT4 is that it’s multi-modal. The interface can accept not only text but images as well. It also increases the number of words that can be input (eight times as many as the original version of ChatGPT), and according to OpenAI, it’s better at tasks that require creativity or advanced reasoning. Being the only free-to-use AI system, its popularity has skyrocketed, and the level of understanding that this engine has is truly astounding, even to those of us who have seen endless forms of AI across various businesses.
How AI Is Impacting Search
Does the proliferation of AI-driven chatbots and answer engines mean that traditional search engines are done for? Are we now in the era of conversational, answer-based systems? That depends on how much the behavior of searching for information will continue to change.
AI will undoubtedly continue to influence search habits, but it won’t squeeze out search engines all together—at least not yet.
AI is moving search behaviors from fact-finding to more of a Q&A, as if we’re talking to a much more intelligent friend or coworker. We ask questions—sometimes very complicated ones—and we can mostly expect to get accurate answers from these engines, although we are not at the point where we can trust every answer – and maybe we never will be. Rather than providing access to firsthand sources of information, AI creates answers from information it happens to have within its own database at that point in time. That’s a different level of trust.
There Are Limitations to Answer Engines
Although being used by major corporations like Microsoft, ChatGPT has been banned in some schools, which can limit its penetration and thus its exposure and ability to learn among those groups.
AI-based systems can also become confused by complicated requests, leading to unreliable “hallucinated” responses (more on that shortly). There will always be lag time between the system’s answers and the most recent database updates, so the engines often answer with outdated information. When looking for quick answers, this could be misleading. (If you dig deeper and challenge the engines, they will often admit to errors in responses based on a lack of current information available to them.)
One of the main barriers to answer engines taking down search engines is the aforementioned hallucination factor. A hallucination is essentially a manufactured response by an AI engine that doesn’t stem from its training input and is created by the AI seemingly out of thin air. Keep in mind that engines like GPT4 don’t scan web pages to find information; they use their base of knowledge to construct answers that they think will best answer your questions. We expect hallucination rates to improve over time, but they’ll probably always be a factor.
Where Search Engines May Have an Edge
Let’s start with Bing, which recently launched an AI-powered version of its search engine. As mentioned, there have been concerns about GPT4 hallucination rates and not providing verified answers. Bing’s new search engine is AI-fueled and feels very conversational, but it also includes citations showing where the information is from and allows users to click through for more information. (Let’s note the wiki-esque risk there: Whether people will actually click through to fact-check, or simply trust the AI answers they see.)
Google recently made its way into the AI game with Bard. In this case, we have an independent AI system created by the most dominant search engine, and it presents a potential argument against hallucination risks as, according to Google, their system is, and always has been, based on real-time information rooted in facts and legitimate data sources.
One of the biggest differences is that Bard is able to draw responses directly from the internet, which allows for more up-to-date answers than ChatGPT. Bard will also create several different responses to every prompt, tarnishing the feel of a conversation in favor of allowing the user to pick the best response.
Bard is currently being tested in the Google Messages smartphone app, giving users the ability to have responses (up to several sentences long) crafted on their behalf which they can review before sending.
So far, Bing looks to be winning the reignited search engine wars. You may notice AI being included in other Microsoft products, such as Teams. Hoping to be known as a main driver of AI, Microsoft is making every effort to turn the tables and pose a real threat to Google.
But search engines need to look out for a new competitor: AI answer engines. (Not necessarily AI itself, as search engines are using AI as an enhancement to their platforms but are still rooted in information-based search).
4 Things AI Could Mean for Marketers
Create Efficiencies. AI could write email subject lines (or full emails), blog posts and articles. It can help answer questions and solve problems that we deal with all day long. It comes down to how much faith there is in these systems to complete these tasks without human input.
Enhance SEO and paid search strategies. AI systems like GPT4 serve as a second layer of a search marketing program. When it comes to systems like Bing’s AI search engine, think about how your organic and paid search listings appear, how they might align with AI-generated responses and if they’re as relevant (and conversational) as they could be, knowing the nature of today’s question-based search behaviors.
Keep an eye out for paid opportunities. If you’re wondering when we’ll have paid ads and sponsored content in AI chatbots, well, we’re already there. Microsoft Advertising just announced a new API that allows companies to deliver ads through chat (although there are still many unanswered questions about it). And it won’t stop there; we’ve just scratched the surface of paid opportunities in this space. But with AI that is search engine–backed and generates ad revenue already, how can we ensure consumers know what’s an ad, what’s factual, what’s a deepfake, etc.?
Don’t worry – AI is not coming after your job. It will supplement the work we do, and we should prepare for its continued growth and impact on our industry, but when it comes to creativity, speculative questions, original thoughts and opinions, AI can’t beat humans.
In the short term, AI could very well streamline lots of information gathering and cut-and-dry tasks that are rooted in doing rather than thinking or feeling.
But, given that AI is artificial by nature, we don’t expect it to replace human input, intellect and ingenuity any time soon. Take our recent work for the Ohio Lottery to support our “Winners” campaign, designed to create awareness of the company’s philanthropic efforts. An AI system could probably come up with a headline like “See how the Ohio Lottery supports education.” But we’re certain it couldn’t produce “Ohio Wins When Ohioans Win,” as our team did.
