8 Comments
Jun 28, 2023·edited Jun 28, 2023Liked by Varun Shenoy

I think the point of ChatGPT is that it was able to do the highly complex task of responding to an empty chat box with a degree of coherence previously not possible.

Yes much better products will come from optimized constraints and context. But the base level of innovation can't be negated nore can we ignore it's rapid adoption per a V1 of Ai interface.

Expand full comment
author

Agreed! I think many models, both closed and open source, can complete highly complex tasks in response to a text input. The question now becomes how can we imagine novel LLM experiences, beyond just ChatGPT.

Expand full comment
Jun 27, 2023Liked by Varun Shenoy

we talked a lot about this in my hci class last quarter actually

Expand full comment

Awesome piece! How do you think about the long-term effects of those limits introduced by somehow predefined options?

Could they potentially hinder the development of increasing intersubjectivity once we move towards more intelligent models that would actually allow for more human:human like conversations?

Expand full comment
author

The interfaces will have to change as the models improve, but I also think over time prompting will get easier *and* models will get better at understanding human language + context. Software is malleable and will have to update as the tech evolves.

Expand full comment

That's exactly the reasoning that made us to build https://www.syntheticusers.com/

Expand full comment

Great thought Varun!

On the complete other end of the spectrum of engaging with data is BI dashboards. BI dashboards are valuable both because they summarize data well but also because they guide the viewer on "how to engage" with the data.

Similarly, I think as we move towards specificity in our representation of the world - we should see interfaces that'll be more opinionated. The specificity maybe a result of narrowing down the domain (e.g. agents to book flights), extracting data better from unstructured formats (e.g. PDF extraction), etc. We are already starting to see some version of this.

Expand full comment

Intriguing insights, Varun! The limitless potential, yet inherent challenge, of controlling and optimizing language prompts in large language models (LLMs) like ChatGPT brings into focus the complexities we often overlook when interacting with AI systems. Similar challenges arise in other areas of our digital lives too. For instance, crafting an appealing online dating profile is a task requiring its own form of "prompt engineering." Much like figuring out the right input for an LLM, it takes skill and finesse to curate a dating profile that accurately represents us and catches the attention of potential partners. To bridge this gap, some innovative services have emerged. One such service, TikTak Studio's AI-Dating Image Generation, leverages AI algorithms to create lifelike images to boost online dating profiles, thereby elevating their appeal and effectiveness. It offers a fascinating glimpse into how AI can simplify and enhance seemingly unrelated aspects of our digital lives. So, while we continue to explore the frontiers of AI in natural language understanding and generation, it's worth checking out how it's shaping our social interactions online. Visit http://dating.tiktak-studio.com to see how AI is transforming the online dating scene.

Expand full comment