Leading Design Teams in the Age of AI

Let's dig into how you can start learning and leading more proactively to support your team as we go through this industry-wide shift.

Hello! Here’s your weekly deep dive into a topic that will help you lead better. Hope you enjoy the read and feel free to forward this along!

At Grammarly, we’re working on new, innovative experiences that push the boundaries of what our Core product is capable of, especially when backed by LLMs. As we’re going through a pretty transformational time in the tech industry, I started thinking about what this shift means for designers and design leaders. How do we show up for our teams to instill a culture of curiosity and fast-paced experimentation, all while upholding a high bar for quality? Although the underlying principles of leadership stay the same, a certain mindset shift is needed when leading a design team in this rather ambiguous, unknown world of AI. Here are a few things that you can start doing with your team to help them navigate the unknown.

•••

Cultivate a culture of learning

AI is undoubtedly the biggest technological advancement of our time. This technology will fundamentally change how we live, work, and operate. If you and your team are new to the GenAI world, the best thing that you can do right now is to instill a culture of continuous learning in your team. A great place to start would be to carve out some time in your calendar each week to take a new tool out for a spin. We’ve been playing around with tools like ChatGPT, Midjourney, Dall•E for some time now and new tools are emerging on a weekly basis (I’m pretty impressed by Sora by OpenAI). Getting to experience some of these tools first-hand has been transformational for me and my team. We have been building a healthy habit around using some of these new tools and sharing learnings as a team. Apart from helping us level up our knowledge within the GenAI space, this ongoing exercise also helps bring the team closer, making collaboration stronger between designers.

Some of the top industries that GenAI is impacting (this list will continue to grow with more use cases):

• Design & imagery — Designers and artists are looking to use GenAI tools to create interesting and useful visuals. These explorations are pushing the boundaries of what’s possible.

  👉 Tools to check out: Midjourney, Dall•E, Visual Electric, DreamStudio

• Content creation — Writers, marketers, and content creators are using GenAI tools to automate tasks and workflows that save them time to take on more strategic tasks.

  👉 Tools to check out: Grammarly, Writer, Jasper, Synthesia, Copy.ai, Podcastle, Murf

• Coding & software development — There are tons of GenAI tools out there that are helping developers (and non-developers) generate code, improve QA processes (huge time saver!) resulting in higher quality output and lowering the barrier to entry for non-developers.

  👉 Tools to check out: OpenAI Codex, Tabnine, Durable, Replit, Github Copilot

• Healthcare — This use case is the most fascinating to me. AI tools are helping medical practitioners accurately analyze images, diagnose diseases, and as a result, predicting outcomes for patients. Radiologists using AI technologies for image analysis reported a 20% improvement in accuracy in detecting subtle anomalies, leading to more timely and accurate diagnoses. I found this article to be pretty informative.

• Finance — Financial firms are starting to use GenAI to analyze market trends, optimize trading strategies, predict stock movements with an impressive 85% accuracy (!), risk assessment and management, and financial planning. A good read on the topic.

Other interesting tools:

Rewind

Glean

Perplexity

Claude by Anthropic

More tools are added to Product Hunt daily

•••

Help your team understand the technical capabilities and constraints of GenAI

If you’re just getting started or have folks on your team who are starting to dip their toes into the GenAI world, this is a good read. There are a lot of unknowns and open questions on what the technology is capable of and what some of the constraints are when designing and building with AI models. A few of my observations (in broad strokes):

• Training data is a requirement — generative models require tons of data to learn from. They’re not yet at the point where they can learn independently. So, the quality of the output that you get is only as good as the data you feed into your models. Richer the data, better the output.

• Not personalized — you may have noticed this when trying out new tools, LLM generated responses are highly generic and not personalized at all. The technology will slowly get there but right now, there’s still you can do to personalize responses to your voice, style, preferences. OpenAI launched ‘custom instructions for ChatGPT’ last year and I’d be curious to see how they evolve this capability further.

• AI will need to get more ethical — this article does a good job of outlining what building ethical AI means and highlights why building ethical AI is important: “…For instance, Los Angeles is suing IBM for allegedly misappropriating data it collected with its ubiquitous weather app. Optum is being investigated by regulators for creating an algorithm that allegedly recommended that doctors and nurses pay more attention to white patients than to sicker black patients. Goldman Sachs is being investigated by regulators for using an AI algorithm that allegedly discriminated against women by granting larger credit limits to men than women on their Apple cards. Facebook infamously granted Cambridge Analytica, a political firm, access to the personal data of more than 50 million users.”

•••

Re-evaluate Design’s role

If your team is designing and building for AI, it’s likely that the way you think about shipping products will fundamentally need to change. You might find your teams wanting to take more of an iterative approach than before. Perhaps there’s a need to move a lot faster than you had been. Truth be told, rapid experimentation may be the name of the game until we find more steady grounds. Here’s the thing about that, though — speed and quality are typically concepts that are at odds with one another. To go fast, you must compromise on quality and to ship high quality products, you must have the luxury of time. This is a fascinating topic, I’ll cover this at length next week. So, stay tuned!

•••

Three things that caught my eye this week:

▶️ Designing AI Experiences

🤯 GPTs: You can create custom versions of ChatGPT

😍 Web design inspiration

•••

Until next time! 👋