Menu
SearchGPT is poised to be the latest and greatest search engine offering available to the public in an AI industry that lives on the cutting edge.
Developed by OpenAI, this prototype is billed as the best of both worlds: combining the conversational tone of AI chatbots with the technology's impressive ability to draw information from a wide array of internet sources. The result is, theoretically, a better and more efficient way for the average internet user to search the web.
Currently, SearchGPT is still being tested by a small group of users. While it is being polished and readied for mass adoption, it is worth a discussion on the efficacy of using AI on such a large and unruly mass of data as it exists on the internet.
One of OpenAI’s goals with this new iteration of search engine is to give users “timely answers with clear and relevant sources.” One does not doubt the sophistication of ChatGPT. However, the sophistication and reliability of the innumerable online sources available on any given topic is a different issue. How are reliable and relevant sources determined?
This question might be of little consequence for menial searches like cooking recipes or bits of trivia, but what about searches where the findings could have a significant impact on a business, organization, or industry?
When the stakes are this high, the datasets are just as important as the tool used to analyze them. Decision makers need more specialized data than a mass search of everything a search engine has to offer. They need data they can trust to deliver insights that can directly impact their bottom line.
How can you trust data, or build datasets that are reliable? Like anything worth scrutinizing, you look at how the thing is built.
The internet is a mixed bag of data because it is an amalgamation of forums, news articles, social media posts, and corporate and personal blogs, among other sites. Undoubtedly, crucial and impactful information is being published every day. The issue is sifting through mounds of useless, if not entirely fabricated, content to find that needle in the haystack that can help position an organization for success.
AI can be much more effective by siloing data into more specialized groups and vetting it for relevance to what is important to a given organization. Get rid of most of the hay, and the needles are suddenly much more visible.
At EdgeTheory, we have adopted this approach. Our datasets are composed of highly specific but exhaustive sources that give key insights into what moves the needle in a given industry. We have developed executive-level narrative briefs that present in-depth data visualizations of the trends decision-makers need to know about.
Built on AI-native technology, we give decision advantage not by using what a generic Google search would render, but by gathering information from both industry-leading publications and fringe outliers who could be sabotaging important narratives to your organization.
By tailoring the data to an organization’s specific area of focus, our narrative briefs help you make better decisions faster.
An important distinguishment between EdgeTheory’s AI and chatbots like SearchGPT is that we make no claims to the veracity of the narratives. In short, we do not present the story as true, only what the storytellers are saying is true. For one, this avoids issues (such as with chatbots) of completely fabricated narratives or misconstruing what is blatantly false to be true.
But on another level, this allows the decision-maker to see the range of how a narrative is being spun by all parties involved. With this context, organizations can be certain where the narratives are coming from, how their messaging is aligned or misaligned with these narratives, and where they can take action to stake their claim in the information environment.
This is another advantage of specialized and segmented data– sorting sources by thematic relevance or inherent bias creates an easier path for analysis and getting a quick and insightful read on what seems to be a complex narrative.
SearchGPT and other iterations of chatbots are great tools for general use online. However, even OpenAI has sought to license vetted content for training its models in order to make search results more reliable.
The size and focus of the dataset matter. It matters even more when the data contains narratives that can make or break an organization’s core mission. When it comes to data, sometimes less is more.
