Sree's Non-Scary Guide to AI

By Janus Boye

Sree Sreenivasan is Co-founder, Digimentors and also former Chief Digital Officer of NYC, Met Museum, Columbia and currently serves on the Nobel Prize Outreach board

“Bullets cannot be recalled. They cannot be uninvented. But they can be taken out of the gun.“

Novelist Martin Amis, who passed away this year, was quoted by Sree Sreenivasan in a recent member's call. Sree discussed the usefulness of generative AI, including its application to our work and the fear it can evoke.

Little doubt remains that generative AI will have substantial significance, but it’s also clear that these are still early days. Will the impact be good or bad? Scary or less scary?

It’s a fast moving field with many big announcements happening weekly, so tuning into a curious and experienced mind like Sree proved quite interesting as always.

Sree is the former Chief Digital Officer of New York City, the Metropolitan Museum, and Columbia University. As he said in the call, he’s been teaching how to use the Web since 1998. Currently he serves on the board of the Nobel Prize Outreach board and he works as co-founder at Digimentors, a digital media agency.

Below you’ll find my summary of the call and towards the end you can also find a collection of useful links for additional reading and you can even download the slides or view the recording.

In a confusing time, Sree opened by setting things in perspective.

ChatGPT can be scary

Sree used this cartoon to illustrate how scary climate change is and how we can expect more hot summers. Similarly with ChatGPT, see below

Sree reminded us that AI features are already in spell checkers, Google Maps, and Grammarly. Then he jumped ahead to the end of 2022 when ChatGPT was released. 

This year, we have seen amazing progress in this field. Sree brought a handful of references from a year full of groundbreaking news:

ChatGPT will get more advanced and more scary - quickly!

There was more AI news from this year in his deck - refer to the slides via link towards the end.

Illustrating his point with a memorable Simpsons cartoon, he made the parallel to climate change and 2023 having been the hottest summer ever on record. Turning that around, he made the eloquent point that if you look forward, 2023 is likely to be considered a cold summer in the future and similarly ChatGPT is the least scary right now.

Let’s take a step back from this whirlwind and look at what’s happening.

Going through the three waves of AI evolution

Sree cited Google DeepMind cofounder Mustafa Suleyman for defining three waves of AI evolution:

  1. Classification: Training computers to classify various types of data like images and text

  2. Generative: The current wave, which takes input to generate new data. ChatGPT is the best example of this

  3. Interactive: The next wave, where an AI will be capable of communicating and operating autonomously.

So what are the things everyone should know about generative AI? Sree illustrated this by showing how popular tools like ChatGPT, Bard and monica.im responded to different prompts with this question. ChatGPT was asked to share the response as a tweet, while Google’s Bard gave a nicely phrased list. Monica.im works as your AI-Powered copilot in the Chrome browser and offers an easy to use interface as a layer on top of the usual prompts.

He also showed an example of DALL-E doing impressionist oil paintings of someone teaching an AI workshop and then showed how it would look differently as a cyberpunk.

In this generative phase we are in right now, Sree made it clear that prompt writing will soon become a part of every job, if it hasn’t already.

Sree then moved onto use cases and shared a vast collection.

AI uses cases today

It seems like generative AI is everywhere we go and Sree opened by sharing how Spotify uses AI to improve the listening experience.

For the frequent travellers among us, Sree shared two services: Room Around, an AI travel planner and also Copilot2Trip, your personal travel assistant. Personally I’m not too impressed with the Aarhus advice you can get in Room Around, but give the free services a go yourself.

There’s also Movie Deep Search, which helps you identify the best movie to watch. It’s very different to Google search, or asking ChatGPT. Here’s some example prompts to illustrate how to use it:

  • If you are a film student, you could ask: “Movies released before 1970 surpass our expectations by asking fundamental questions with very clever writing in a smooth flow of a plot” and if you like, you could add more depth by asking it to only search Netflix

  • If you are a film screenwriter how about this: “Well-constructed crime film of traditional narrative along with a good screenplay make it an amazing piece of work” and then move forward say with “strictly the British ones with tight script”

He also shared an Instagram hashtag generator and Pi, your personal adviser. While ChatGPT generates text based on a large database, Pi is focused on having natural and empathic conversations. They call it “personal intelligence”.

Feel like going wild with an angry email? Sree pointed us to AngryEmailTranslator.com, which will help you make it work-appropriate as shown in the screenshot below.

If you are on LinkedIn and have “Creator mode” on, you might have noticed that LinkedIn now offers writing prompts based on your skills and also helps you flesh out status updates based on short prompts. On the topic of business networking, and perhaps changing jobs, Sree also pointed us to resume.io, a free resume builder.

Worried about whether this or some other text is AI-powered? Then check out AI Content Detector. When I gave it 1,500 words from this text, it came out 99% human-generated, which is actually only 1% off.

Ending on a more scary and ethically questionable note, Sree told us about how AI can make skin tones lighter and change facial features when used to create headshots. Read the WSJ story here: AI Botched Their Headshots. There’s more problems with fake content and Sree closed with a warning for 2024 that connected us back to the opening quote.

Like he said: It took us 60 years to realise the big dangers of tobacco. With oil we are still working on it. Now with less than a year of generative AI, we can see how powerful it is. In other words: How do we take the bullets out of the gen AI gun to avoid enormous problems, also when it comes to the impact on democracy and upcoming elections around the world?

Learn more about what’s happening in this AI moment

At the recently held Boye Aarhus 23 conference, several speakers and conference sessions focused on this emerging topic. Notably the thought-provoking conference keynote by CWI researcher and Internet pioneer Steven Pemberton on There's no I in AI (view slides as HTML).

Sree hasn’t yet made it to one of our in-person events, but he did also do a popular member’s call with us during the first wave of the pandemic in May 2020. Read: The World Has Changed. So How Are You Going To Change? Since before then, I’ve subscribed to his excellent newsletter called Sree’s Sunday Notes.

For some additional reading on how AI changes everything, here’s just a few pointers from other corners of our community:

Towards the end of the call with Sree, he shared three of his essays on the topic:

You can also download the slides (PDF) as they were used in the call or view the updated version (Google Slides) which Sree continues to update.

The conversation naturally continues in our peer groups, which meets every quarter in Europe and North America. There’s also plenty of AI at our upcoming 2024 conferences, including:

Finally, lean back and enjoy the entire recording below.