4 Ways How ChatGPT (GPT-4) Improves B2B Content
A month ago, we discussed the positives of AI writers (including ChatGPT) for B2B content and research. The focus was on outlining how GPT-3 and GPT-3.5-powered applications could help with: Providing writing inspiration Writing templatized content (like SEO titles and meta descriptions) Creating Lo-fi content outlines More about it here. Within a month, GPT-3 and GPT-3.5 have become legacy models – thanks to the advent and widespread adoption of GPT-4. On March 14, 2023, OpenAI announced GPT-4. A model that could outperform the “English-language performance of GPT-3.5 and other LLMs (Chinchilla, PaLM)” in 24 out of 26 languages A model that could score 163/170 (GRE Quant) and 169/170 (GRE Verbal) as compared to GPT-3.5’s 147/170 and 154/170, respectively Most importantly, OpenAI claimed that GPT-4 was 40% more accurate in producing factual responses than its predecessors. Of course, this raises a few questions: Does this mean that GPT-4 hallucinates less? How relevant are the responses when the subject is complex? Can the improved capabilities help with B2B tech content writing? We set out to understand these by using ChatGPT Plus, which enables you to elicit responses from the GPT-4 language model. Here are the things that make it an excellent resource for B2B writing. 1. Building On One’s Own Research There’s this trend of creating content from scratch using ChatGPT. While that isn’t against Google’s guidelines per se, there’s a huge possibility that the final write-up would end up generic, uninspiring, and lifeless. Why’s that the case? For two major reasons: AI’s controlling the narrative Linguistically inferior content When AI’s controlling the narrative, you overlook the in-house insights that would have otherwise helped differentiate the article while still keeping in line with what the consumers need. Plus, you get little opportunity to share your experiences and expertise – components highly valued by Google’s search algorithms. About linguistically inferior content – this has a lot to do with conceptual redundancy and bland articulation. There’s enough literature on the latter, and a host of AI detection tools have also graced the market in line with the same. These tools validate the humaneness of content by underlining its “randomness” and “perplexity” — a practice highly revered in academia. But what’s conceptual redundancy? It’s the repetition of ideas. Often, AI repeats the same ideas while tweaking the language to make it sound authentic. From a reader’s perspective, that doesn’t bode well for engagement and consuming value. This problem is easy to solve with ChatGPT Plus (when eliciting responses from GPT-4). For example, if you want to get some inspiration around the research you’ve done for creating an outline for a blog, ChatGPT Plus can help build on top of that research. In the following conversation, I asked ChatGPT to build on top of my own research about the use cases of generative AI in healthcare. Medical image analysis, remote patient monitoring, and mental health chatbots aren’t novel concepts. However, “Diagnostic Support” and “Personalized Treatment Plans” could have a novel angle to them. So, I probed ChatGPT further and asked it to restrict its research to generative AI capabilities and not go down the path of elaborating on core data analysis and machine learning concepts. These are excellent responses because they tell me how generative AI technology could work through end-to-end streamlining of a healthcare service provider-patient interaction. At the end of the day, starting with an established narrative serves better to elicit profound responses from AI. These responses: Aren’t exceptionally generic Are in line with the narrative’s vision Open avenues to think about novel applications Also Read: Can AI Write B2B How-To Guides? Findings From Our Experiment Suggest Otherwise 2. Conceptualizing the Correlations Often in B2B writing, the subject matter is complex. That’s because it harbors several inter-relationships and nuances. So, how do you effectively elucidate these concepts? One of the ways is to quantify or visualize the correlations between various variables. For example, you can be talking about the market’s preference for microservices over monolithic applications and the corresponding shift in cloud consumption patterns. Likewise, you can be talking about the overall decline in traditional advertising channels (e.g., TV, radio, print, etc.) and the corresponding shift in the use of first- and third-party data to support dynamic or programmatic marketing. When you’re dealing with such abstractions, which are almost always engaging if they come out well, it’s viable to let AI assist you with content ideation. Such ideation opens avenues to think more granularly, come up with captivating writing ideas, and also get a peek (albeit assumptive) into what readers would find engaging. ChatGPT is really good at brainstorming through correlation. For example, you can ask it to connect two seemingly disparate concepts to better elaborate the issue. Here’s a conversation I had with ChatGPT Plus when I wanted to illustrate the overall structural trend of green initiatives. Media coverage, CSR, supply chain sustainability, financial performance, etc., are excellent correlations to elaborate and visualize based on real research data. So, this conversation with GPT-4 went well to build on top of the starting point and broaden the scope of ideas while keeping the focus narrow. Also Read: 4 Areas Where AI Writers Fall Short for B2B Tech Content Writing 3. Including Highly-Relevant Examples A similar use case as the one above is providing industry-specific examples to support your ideas. Ideally, examples serve as credible and objective evidence for your argument. They help map a process with a corresponding outcome, a product with its real-life use case, and a service with its utility. In the B2B domain, in particular, examples bridge the abstraction gap and prove the value of a service or product. The key, however, is to choose relevant examples that 1) work well with the underlying narrative and 2) are relevant to your target audience. Again, ChatGPT Plus can be prompted to come up with relevant examples to support your research. These examples aren’t just viable for B2B long-form content but also explainer videos, social media posts, and other marketing collateral. Remember,