Building AI features while maintaining a balance between their needs and aspirations
Since every company is rolling out their AI feature, every product is powered by AI. As part of the build team for our product, I feel a strong responsibility to join this trend. But I don’t want to fall into the trap of creating a generic rewrite tool or a basic feature. I want to build something that truly helps users. At the same time, I don’t want to set false expectations, promising magical results that we can’t deliver right away.
I realized that the majority of AI features on which the companies are currently working are divided in two. One is need basis, like an assistant, and the other is the magic phenomenon. So, in the need, in the assistant phase, it usually solves the user's needs. Examples are rewrites, quick search, interpretation, and chatbots. The other is the aspirational one, which makes you think that this work would take you another skill to learn and hence solve the skill gap. But this is what AI is doing for you. So, it does a magic.
How people are using GenAI
I would look into how people use AI to understand how to achieve this balance. I have discovered that AI is mainly used for:
Troubleshooting problems and technical assistance
Reading, writing, and editing content
Personal and professional support as an assistant
Learning new things
Users typically start by explaining their situation. They might detail where their product is live, the specific problem they face, who their users are, and share relevant screenshots or notes. Then, they ask the AI for specific outputs, like generating marketing headlines or rewriting text. They provide details about the tone, style, and constraints, such as requesting a friendly tone, active voice, or character limit.
After receiving the AI’s output, they iterate, comparing it with their past work or inspirations, refining it until they achieve satisfactory results. This pattern of providing context, asking specific questions, defining constraints and rules, giving examples, and iterating is clear.
Exploring Competitors and Broader AI Trends
With this insight, I would explore how other companies, both direct competitors and everyone broadly in the AI space, are developing their products. I would quickly categorize their AI features into need-based and aspirational. However, I would still need a clear direction on how to proceed. So, I would research current AI usage and iteration patterns further.
The next product to be built should strike a balance between being an assistant (meeting immediate needs) and providing aspirational features (long-term goals). I don’t want to offer just basic functionality, which should already exist, nor do I want to overpromise and underdeliver, which could erode user’s trust.
Identifying User Problems and helping them grow
Next, I would focus on understanding the specific problems users face and how AI can help them. I aim to find issues users struggle with and see how AI can simplify their tasks. To achieve this, I would learn from real users about their workflows, what they are doing, and how they give context and ask specific questions. Users typically provide context, ask for specific outputs, define constraints and rules, give examples, and iterate.
After gathering this information, I would train our AI model using feedback from power users. These users provide insights into the most needed features and adjustments. For instance, if users frequently ask to change the tone or make text simpler, I would add a “tone reading level” option. Additionally, I would include basic options for adding constraints, rules, and examples. This approach aims to make the AI more intuitive and personalised, addressing common user needs directly and solving their issues.
Ensuring Data collection is ethical
Before rolling out the AI feature publicly, I would want to ensure it is ethical and doesn’t breach user’s privacy.
Collecting data is crucial for training the model, but it must be done responsibly.
Therefore, I would ask for user’s permission first, explaining how their data would be used and ensuring they are comfortable with participating in the beta test. Only those who agree would be included, ensuring transparency and respect for user consent. This step is vital to build trust and demonstrate our commitment to ethical practices.
In addition to seeking user permission, I would follow public ethical guidelines to make sure our practices align with industry standards. These guidelines cover various aspects, including data privacy, security, and ethical AI usage. By following the standards and guidelines of the industry, I aim to prevent any potential misuse of user data and ensure the AI’s operation is fair and responsible. Once I am confident that all ethical considerations are met, and the AI feature is both effective and trustworthy, I will proceed with delivering the product. This thorough approach helps maintain user trust and ensures the AI is developed and deployed responsibly.
Conclusion
Building an AI feature that balances need-based functionality with aspirational goals requires understanding user needs and ensuring ethical practices. By exploring how people currently use AI, researching competitors, identifying specific user problems, and training the model with power user feedback, I aim to create a product that truly helps users.
By also seeking user permission and following ethical guidelines, I ensure the AI feature is trustworthy and responsible. This process would help balance immediate user needs with long-term aspirations, while maintaining trust and ethical standards, to build a successful AI feature.
