You wouldn’t fault an explorer for their love of the compass.
Innovative technologies make it hard to limit our excitement. When something comes along that has a radical impact on how we work or interact with our world, we want to embrace it.
Sometimes it’s best to take a beat to dig in a bit more, even when we’d like to think some new tool or invention is a cure-all. Without understanding the nuances and implications, making that head-first slide into home plate might be riskier than we thought.
The age of generative artificial intelligence (AI) is here. As we move into this new chapter of human history, the race is on to see how to adopt this new technology. In this article, we’ll explore how generative AI applies to B2B software reviews and G2’s stance on it.
Innovation at breakneck speeds
Generative AI tools like DALL-E 2 and ChatGPT first made headlines in late 2022. Just a few months into 2023, adoption and interest in AI technology have exploded. With more software vendors exploring how to integrate AI to create new solutions, professionals across industries are hailing this as a new frontier for businesses.
Some argue that generative AI has the potential to boost global GDP by 7% by 2030. On the flip side, many foresee how this technology might impact the global labor market. What the extent will be remains to be seen, but experts suggest the likelihood of it changing how we work very soon.
Although most notably known for generating text and imagery outputs, generative AI has the potential to do other things like write code and help with data analysis. In one iteration of ChatGPT, a user successfully used this tool to create a playable version of Pong in under 60 seconds.
Generative AI has far-reaching implications, regardless of industry. But because this technology is making advances at such a rapid pace, it also presents unique challenges and potential for abuse by bad actors. Before making the connection between this innovative technology and G2’s policies and processes related to review moderation, it’s important to establish context about generative AI.
Challenges of exciting technology in early phases
AI-generated content is set to revolutionize content creation in the coming years and likely disrupt existing creative industries. Because this technology is brimming with such potential, there’s also a massive possibility of widespread abuse.
There’s no denying that generative AI is impressive. However, consider the following challenges and risks before taking the dive.
- Ethical and copyright concerns: Many opponents of AI-generated content note the ethical concerns around it. Since these models train on existing content, citing sources and giving credit becomes complex. From college students using AI to write term papers to "borrowing" code on a massive scale, the risk of plagiarism and copyright infringement are very real.
- Content quality: Generally speaking, generative AI tools are compelling. But as any writer or artist will tell you, they won’t replace human jobs anytime soon. The future of AI will likely be a combination of technology and human operators but don’t expect to replace your creative team with these tools just yet.
- Misinformation and inaccuracies: In a similar vein to content quality, generative AI has little in the way of preventing inaccurate, false, or misleading outputs. Publishing inaccurate content can damage a brand’s reputation, so it’s crucial to remain vigilant about fact-checking AI-generated content.
- Learned biases: Another aspect raised by critics of AI-generated content is that their models use human content. The argument is that because humans have implicit and explicit biases, these AI models learn these patterns and behaviors.
Our stance on generative AI and software reviews
With over 2M+ reviews from real folks, G2 is the world’s largest marketplace for B2B software. Getting to this point required a ton of hard work and dedication from our team. But it wouldn’t be possible without respecting the power of unbiased, authentic reviews.
At the heart of it all, we know that humans trust the experiences of others, which is why we prioritize the quality of reviews over quantity. Buyers have come to trust G2 for researching and evaluating software because we do everything we can to guarantee authenticity, such as:
- Verifying the identity of every user on G2.
- Leveraging over 43 data points to assess authenticity.
- Vetting every review using strict guidelines.
G2 takes an unbiased approach to moderate reviews. Our goal is to guarantee every review is the honest opinion of the reviewer’s experience with a product. We also actively monitor for patterns that indicate fraudulent behavior. When such instances occur, we perform a thorough audit, takedown the reviews in question, and implement necessary enforcements.
Because of our commitment to unbiased buyer feedback, reviews that use generative AI to write a majority of the content do not meet G2’s community guidelines.
While we remain firm that reviews should come from real folks, the one exception is using AI writing assistants. So long as the tool only aims to improve readability, grammar, and spelling, we welcome reviewers to utilize these tools.
It’s also worth noting that G2 is carefully considering the benefits of incorporating AI tools into the review submission experience. Similar to how writing assistants can help reviewers, we’re constantly improving the overall submission process.
How G2 review moderation detects AI-generated content
Our team has a rigorous vetting and auditing process for all new review submissions. Because of the massive number of reviews we collect on any given day, we must use a thoughtful blend of best-in-class technology and well-trained team members with a discerning eye.
4K+
AI-generated reviews G2 rejects per month, accounting for 5-8% of total monthly review submissions.
G2 uses leading third-party tools and services to detect AI-generated content in review submissions. While we won’t tip our hand on the specific technologies leveraged, rest assured that we utilize some of the best-reviewed software on the market.
The tools at G2’s disposal are enterprise-grade content detection solutions that scan and flag AI-generated content. After flagging this content, our in-house team of qualitative moderators validates these instances.
Good intentions vs. bad actors
It’s important to acknowledge that we discourage and reject submissions that use generative AI. Because of our stance on the power of authentic reviews, we aim to decline reviews when our processes detect AI-generated content, regardless of whether the intent was legitimate or fraudulent.
In addition to using third-party tools, we utilize an internal ML model that learns from each moderator's decision to detect fraudulent reviews. This ML model also uses other data, such as a user’s review history, social media presence, submission pattern, browser actions, IP addresses, location information, and more to detect fraudulent activity.
And while AI content detection and our ML model help flag fake reviews at scale, we also layer in the specialized expertise of our moderators to audit reviews that require more considerable evaluation. Our team is adept at identifying patterns commonly associated with fraudulent activity and creating effective fraud prevention practices.
Laying the groundwork for a brighter future with AI
There’s a good chance that the world will look much different in the coming years due to the explosion of generative AI technology. Countless content creators and brands will have to adapt and master these tools to stay competitive.
But for as much good as this technology can do, the same can be said for the harms it could potentially wield. G2 embraces and recognizes the tremendous upside generative AI presents. However, we still feel that reviews are only trustworthy when they come from the first-hand experiences of real people.
Because we believe in the power of authentic reviews, you can always count on G2 to champion the voice of the customer.
Find the right software with the power of AI. Explore a new way to research with G2’s AI-powered chatbot Monty.
Meet your new software buying assistant.
by Eunice Buhler
by Soundarya Jayaraman
by Andrew Stevens