What would be an appropriate task for using generative AI?

Generative AI is best for content creation, summarization, brainstorming tasks, coding support or help (writing code), translating languages, and data analysis. Please provide an instruction and some context to start from. It's of limited utility for many things that require up-to-the-minute information, deep subject-matter expertise, or human-in-the-loop decision-making.

Generative AI is long past the peak hype cycle. Now, millions of people rely each day on tools like ChatGPT, Claude, Gemini, and Jasper—but many don't know where AI truly helps and where it only disappoints.

That gap matters. Most organizations find generative AI failing at tasks it isn't designed for, with robots outputting noise instead of results and misusing vast resources. Measurable productivity gains are made by those that do. Generative AI can generate $2.6 to $4.4 trillion in annual value across industries, but only with proper deployment, according to a recent McKinsey report (April 2023) [7].

The question is not whether you should use generative AI. It's also known as "How do I make use of it?

In this post, we'll explore where generative AI reliably pays off and when you should proceed with caution—as well as how to tell the two apart. By the end, you will have a working framework to make better decisions on how AI fits in your process.

What is a task that is good for generative AI?

It helps to understand what generative AI does well, given its general structure, before getting into specific use cases.

Generative AI models are trained on mammoth sets of human-generated text, code, and other media. That's what makes them so good at.

• Pattern Recognition: Learning from Examples and Inferring Structure and Style

• Generating language: Generating natural, coherent text from prompts

• Synthesis: Combining information from multiple sources into a single output.

• Iteration: Having lots of different takes on the same piece of content, in short order.

The tasks that fit neatly into this rubric generally have a few things in common: they make up the creation or manipulation of content; they are amenable to speed and scale; and the output is reviewable and improvable by a human.

What content creation tasks can be effectively performed by generative AI?

Generative AI has left its most visible mark in content creation—quite rightly so. AI excels at writing first drafts, reformatting content for other channels, and scaling written output.

First drafts & conquering writer's block

Creating that first empty page is still often the biggest challenge with any writing task. Lay down a brief, and generative AI tools like ChatGPT or Jasper can spit out a usable draft in seconds. This doesn't mean replacing the writer; it means eliminating the friction that slows them down.

Common Applications of AI—Writers, marketers, and content teams use AI for blog post outlines, email campaign drafts, product description creation, and social media copy. The secret is treating AI output as a first draft rather than a final version. That said, human editing is still necessary for accuracy, brand voice, and originality.

Repurposing content across multiple formats

One long-form blog post can be turned into a LinkedIn article, email newsletter, YouTube video script, and social media posts. Read more on: Automated video dubbing is still a way off. How long does it take to dub content into another language manually? Generative AI can handle those transformations in minutes.

This use case is especially useful for marketing teams that have to manage many channels but have a limited headcount. Teams can create platform-specific variations from a single source of truth rather than building every piece from the ground up.

Writing product descriptions at scale

Brands managing thousands of SKUs face a practical issue: how to write unique, enticing descriptions for each product. By taking data inputs such as product specs, category, and target audience, generative AI can generate structured product descriptions at scale. That's why retailers like Zalando and Amazon have already started integrating AI into their product content workflows.

Coding and software development is where generative AI can help.

Generative AI has become a real productivity tool for developers—not just for writing code, but for everything else that slows development cycles.

Generating, explaining, and debugging code

GitHub Copilot, for example, is also driven by OpenAI Codex, and it autocompletes functions, recommends API implementations, and highlights errors in real time. According to a 2022 GitHub study, Copilot users completed tasks 55% faster than programmers working without it.

Beyond writing code, generative AI can also explain what that code does. This is especially helpful for developers new to a codebase or for non-coders who need technical explanations without wading through raw code.

Writing test cases and documentation

Writing tests and keeping documentation up to date are two tasks every developer loathes. It can write unit tests from function definitions and write technical documentation from commented code. Neither output is perfect, but both give developers a solid head start—and sometimes that is all they need to speed up their workflow.

How is generative AI used for research and summarization?

Generative AI reads a greater volume of text more efficiently than any human reader. This makes it perfect for synthesis and summarization tasks!

Summarizing long documents and reports

Generative AI can also distill legal contracts, research articles, financial filings, and meeting transcripts into easily digestible summaries. Claude and Gemini, for example—tools designed to accept long-context inputs—can sift through hundreds of pages to find the most pertinent information.

This saves knowledge workers time when they need to get up to speed before a meeting, client call, or decision.

Brainstorming and idea generation

Think of generative AI as your round-the-clock brainstorming partner. The AI writing tool takes in a problem you want addressed, a product category, or even a few sentences of the creative brief before spitting out dozens of ideas in seconds. Most of them will not be good, but just the sheer amount accelerates the ideation process drastically.

Market researchers utilize AI to investigate customer groups. It helps product teams stress-test product ideas. It gives writers angles they otherwise never would have thought of. The insights aren't valuable on their own; their value comes from how quickly AI can surface possibilities for humans to act on.

Where does generative AI create value in the day-to-day operating model of your business?

Generative AI is moving beyond creative and technical tasks and can now often be embedded in regular business operations.

Customer service and support

Generative AI-powered conversational chat agents can address routine customer questions, escalate nuanced queries to human representatives, and generate customized responses at scale. Using generative AI, tools like Intercom and Zendesk are embedding the technology into their support workflows so teams can close tickets faster without hiring more people.

But beware: AI-assisted support works best for frequently asked, clearly articulated questions. It's not very good at the subtle, emotionally sensitive things that require real empathy and judgement.

Writing about HR and internal communication

These include job descriptions, performance review templates, onboarding materials, and internal policy documents. The documents have a predictable structure and require no creative or artistic language, which is another area where generative AI has performed reliably.

Data analysis support and reporting

In fact, generative AI can turn irregular data into written stories when coupled with the correct integrations. Applications such as Microsoft Copilot for Excel can condense spreadsheet information, highlight patterns, and create simple reports in natural language. This is critical for analysts who need to share their discoveries with non-technical audiences.

Which tasks should you not do using generative AI?

Knowing what AI can do is just as important and necessary as knowing what it cannot.

High-stakes medical, legal, or financial advice—Generative AI can generate feelings of trust and plausibility in these domains but lacks the licensing, accountability, and verification that professional advice provides. Treat AI as a research and drafting aide instead of an end goal.

Most generative AI models have a training data cutoff, meaning they can not access real-time or frequently changing information. They cannot browse the web, so they do not know of events after October 2023 unless they use a retrieval system. Looking up facts through AI without double-checking is bound to be incorrect.

Deep causal reasoning: AI is a pattern-matcher, not a logical reasoner. Sure, they can simulate reasoning to varying degrees of effectiveness and believability — but what involves complex causal analysis (specifically in the case of situations not previously experienced by computer systems) is still not something humans will be replacing.

Sensitive interpersonal correspondence: Performance improvement meetings, conflict resolution, or any other emotionally charged correspondence should be human language because humans with true empathy are better at these conversations than even the most sophisticated AI.

How to decide when and how to use generative AI: A pragmatic framework

• Employ this decision framework to help you assess whether a task is suitable for generative AI:

• Is the job really about producing or converting content? If yes, AI is likely useful.

• Is the result reviewable and vouchsafable by a human? If no, proceed with caution.

• Does the task involve real-time data/expertise? If so, AI should be the helper—not the one taking charge.

Would increasing velocity and volume improve the results? If so, AI has a clear value proposition.

Are the same functions and a clear manner of approach always involved in this task? In that case, AI can alleviate a lot of manual burden.

Do one thing, then do another.

The organizations deriving the most value from generative AI aren't rolling it out everywhere simultaneously. They began with a clear, narrow use case—writing email drafts, summarizing reports, and generating product descriptions—then refined the results using human-in-the-loop validation.

Generative AI is a chance accelerator, not a replacement for human discretion. Those teams that understand both halves of that equation are the ones building sustainable workflows.

Choose a task from this list that hits an existing bottleneck in your workflow. Run a two-week test. Assess output quality and time savings. There is only one experiment-will-classify, which teaches you more about the compatibility of generative AI with your context than any benchmark or case study.

Frequently asked questions

In the day-to-day work, what is generative AI best used for?

Generative AI shines in content generation, summarization, brainstorming, code assistance, and evaluating communication drafts. These are the kinds of jobs AI can do quickly and generate multiple variations in the same language. Is human review still necessary to ensure accuracy and quality?

Is generative AI capable of taking the place of human writers?

While generative AI is very good at quickly drafting content, it will not replace a human writer. It lacks lived experience, true imagination, and the ability to acknowledge the content it generates. Most professionals use AI to speed up drafting and ideation, while humans handle editing, strategy, and final judgment.

Research reliability of generative AI

While generative AI is often effective at summarizing new content, its design can also lead it to fabricate information on narrow, up-to-date topics. Always double-check AI responses to questions against a source of truth for your field when using them for research.

Which industries stand to leverage generative AI the most?

The areas of highest generative AI value across industries, as identified in a 2023 McKinsey summary analysis, are marketing and sales, software engineering, customer operations, and product R&D. However, generative AI is also being rapidly applied in healthcare, legal, finance, education, and retail.

I think about this at work—what is a sufficiently complex task to be beyond the reach of generative AI—and I have not come up with useful, generalizable answers.

Generative AI probably can't perform a task if it requires real-time, rapid access to data; involves serious ramifications (legal or financial); or comes down to subtle human judgment—especially when it requires nuanced communication in an emotional context. For those, AI can help prepare and draft—but should not be the ultimate decision-maker.

Data up to October 2023! CSU training on how good and bad generative AI prompts work. Images are for demonstration purposes only

An effective prompt is specific–it should provide context. The type of format or tone expected. Be well-bounded. Input a generic prompt—you will get a generic output. This would be something like "a 100-word product description of a noise-canceling headphone for remote workers, a friendly and direct tone" that performs better than simply asking it to write a product description.

Disclaimer