Using AI for text summarization

There are often times where marketers need to summarize text. Think of client or customer interviews that have been transcribed but end up in lengthy documents. There could also be lengthy product documentation that needs to be adapted into something much shorter for a website or the blog for a product launch.

Generative AI tools are well-suited to this task and can be a marketer’s friend when it comes to processing lengthy text. Let’s take a look at an example.

Editing a transcript into a simple summary

In this case, we’ll be using this transcript from a 30-minute podcast. Anthropic’s Claude is a useful tool as the chat interface has an attachment button that allows you to upload pdfs (if you don’t already have an account, you can sign up here). All you need to do is use a simple prompt like ‘provide a bullet summary’ with the file that’s uploaded.

Here is the text that Claude provides based on this prompt:

• Kaia was born in Ohio but grew up in Los Angeles. Her family has roots in the rural South. As a child, she spent time on her grandmother’s farm in Louisiana, which was her first exposure to agriculture.
• Kaia has a background in journalism and entertainment media. She leverages her journalism skills as a professor teaching writing and food sustainability courses at NYU.
• While stuck in Italy during the early months of the Covid-19 pandemic, Kaia recognized that food insecurity would become a major issue in the U.S., disproportionately impacting Black communities.
• To address this, Kaia created the Black Farmers Index - an online directory for consumers to connect with and purchase directly from Black farmers and growers. This provides an economic lifeline for the farmers while increasing food access for consumers.
• The Black Farmers Index has grown substantially since its launch in March 2020, now featuring hundreds of farmers and becoming a platform to highlight Black contributions to agriculture.
• Corporations and large institutions wanting to diversify their supply chains struggle to effectively support minority farmers because their business models do not align. Kaia argues they need to create more flexibility and hold space for these farmers to truly uplift them.
• Meaningful diversity, equity and inclusion requires moving past performative efforts and discomfort to enact real systemic change. This involves acknowledging and addressing legacies of racial injustice.

It has pulled out the main points from the lengthy transcript. You can also ask for this in different formats, like as a paragraph, or even in a specific tone of voice (say more conversational, if you need content for a blog).

Interrogating lengthy text

Where this now gets really interesting is that you can ask the AI tool to find specific information from within the document. Here’s an example in response to the question “what is the problem with DEI? source the result from the attached doc”

According to the transcript, Kaia expresses some critiques of the current state of diversity, equity and inclusion (DEI) efforts:

• She states: “I don’t like diversity, equity and inclusion that that term, that phrase because it is it is very performative, as the young folks say, you know, in most areas. That’s why I don’t like it. If you really want it, then you just be it.”
• She argues that many corporations or people have not had to exist in “problem solving spaces” related to racial inequity before. When presented with concrete steps to address these issues, it’s uncomfortable for them because they’ve existed in spaces of “complete privilege” and “entitlement” without accountability.
• She states that meaningful change requires moving past performativity and discomfort to enact real systemic change. This involves acknowledging legacies of racial injustice that companies and individuals have benefitted from or been complicit in.
• Overall, she critiques DEI efforts that are surface-level or for show, without doing the deeper individual and institutional work to address systemic racism and make space for those who have been excluded. Simply stating values around DEI without systemic change is insufficient in her view.

Note that to make sure the tool only looks for an answer within the document, you need to focus the prompt using language like “source the result from the attached doc.” Also, you don’t need to reattach the document each time: it’s sufficient to do all of this communication in the same thread.

One word of caution: do not take text verbatim from a tool like Claude and publish this externally, say to a website or blog. In these instances, think of the tool as a useful input source, but make sure you review the text for inaccuracies and update it to reflect your voice.

As you can see summarization using Generative AI can be a time-saver for marketers and can help us focus on the more strategic aspects of the job.