Arian Shafa

There are studies that show AI use can lead to less collective creativity (see references 1 & 2), while others show how AI is increasing idea diversity (references 3 & 4). So, which one is it?

As with most nuanced things in life, it’s a combination of both — and many other things — and it depends.

It’s a combination of both:

I see these models as an ocean of knowledge, wherein every specialty (e.g., psychology, literature, history, etc.) is but the size of a small pond.

So yes, since almost everybody is using the same ocean when they prompt their AI — putting aside the differences in the model provider, version, harness, context, and prompt — it becomes harder to collectively use the knowledge outside this ocean.

At the same time, the now-connectedness of all these old ponds in our LLM ocean means that at any given moment, our prompt can access, build upon, or analyse more knowledge than humanly possible. Your favorite teacher might’ve had access to a few half-ponds of knowledge relevant to their domain; your favorite professor maybe a few full ponds. And humanity’s brightest minds — Einstein, Claude, Da Vinci — might’ve had access to several ponds. But all the knowledge in an LLM is simply too vast for a single human to process, let alone remember and use when prompted.

And we have access to all of that, with AI, when trying to come up with an original idea or decision.

It depends:

It also depends on how we use these AI tools. Do we form a hypothesis ourselves, and then ask AI only for supportive viewpoints, do we even form a hypothesis, or do we ask straight for an answer to a half-baked question?

When prompted, one suggestion my AI had, based on the Wong & Qiu (2026) paper in reference 5, was to use AI as a cognitive exoskeleton:

“If you let [AI] move for you, your own muscles might get weaker. But if you use it to extend your reach, after you've moved on your own, it can expand the range of what you can do. So a practical takeaway might be: generate first without AI.”

Then expand on your thoughts with AI.

References:

1. de Rooij & Biskjaer (2026) — Does AI make us think alike? (tandfonline.com/doi/ful…)

2. Doshi & Hauser (2024) — Individual creativity versus collective diversity (science.org/doi/10.1126…)

3. Lee & Chung (2024) — AI can enhance creative thinking (nature.com/articles/s41…)

4. Ashkinaze et al. (2025) — AI can increase idea diversity (juliamendelsohn.github.…)

5. Wong & Qiu (2026) — Think independently before using AI (link.springer.com/artic…)

At no point in human history has technology, and therefore our power and control over our environment, moved this much, this fast, for this many people.

My thoughts on the more than one billion daily active ChatGPT users.

“Your body is run by proteins: tiny machines, millions of them, making everything work; from digesting your dinner to fighting off a cold. What each one depends on, is its shape. So, if you want to understand a disease or build a drug, you need its shape. And getting the shape of one protein, just one, can take a scientist years, sometimes a whole career.

Alpha Fold has now done 262 million of them. And he [Demis Hassabis] has given every single one away for free to every researcher on earth.”

I love how succinct this description is.

It's wonderful to hear about all these amazing improvements.

Credits: David Joseph, Royal Society of Arts, Albert Medal Lecture

Sometimes I try to stop and take in all the ways in which my life has been more pleasurable, compared to a year or three ago.

Last time I did this, I realized how immensely grateful I am for AI. I know, it’s not the most politically correct thing to declare publicly, but integrating AI in my decision-making process has increased my quality of life enormously.

Take travelling: I used to spend hours trying to compare prices, locations, and other metrics to ensure I’m planning an okay holiday. Now, with AI, I’m still making the final decisions and validating AI’s assumptions, but I’m having access to a lot more inference/intelligence a lot quicker, which results in a much better overview of everything, better analysis of different options, and essentially less stress and more joy. I’m sure I’m not alone in feeling this way.

This feels eerily similar to the subprime scheme of 2000s, eventually leading to the 2008 financial crisis…

»» https://www.reuters.com/technology/wall-street-giants-partner-with-nvidia-500-billion-ai-financing-deal-ft-reports-2026-08-10/

1.5 trillion leasing/debt, 1 trillion of which doesn't show on the hyper scalers' balance sheets…

»» https://www.ft.com/content/a0a07cce-6d19-4b1e-a73b-9855a06ba7b3

Super excited to become a board member at BuurtBuik: where we fight food waste by repurposing food that would otherwise go to waste and providing it to people in need.

We do this by cooking and repackaging the food we receive from our partners –from supermarkets to hotels to companies that provide their employees lunch/dinner– and serving it to our guests every Saturday.

What personally kept me going back was, among other things, meeting so many lovely people outside my social bubble. It also makes me feel closer to my grandpa, a restaurateur, whom I love very much and who has passed away.

If you want to learn more, volunteer or partner with us, feel free to reach out to me, or read more about the organisation here: https://buurtbuik.nl/

The cherry on top? Not a single person in this organisation is paid: not the volunteers, coordinators, or board members. We have zero personnel costs, which is something I’ve never seen at any of the organisations I’ve volunteered with.

Thank you Joe Llewellyn and the rest of the team for the collaboration so far; onwards and upwards! 🚀

Agentic Engineer Solution #2: Hope

Have you ever felt frustrated that most news is bad news?

To balance the scales, here’s Hope: https://lnkd.in/enywdVh4

You can read many pieces of positive news every day. And on top of that, you can register for the newsletter to get the top 5 most positive news on a daily, weekly, or monthly basis.

All the positive news is also categorised dynamically, meaning today’s categories may differ from tomorrow’s, so that you can find your favorite themes more easily, every day.

Are there any features you would like to see?

Agentic Engineering Solution #3: Our Hopeful Past

With the original website, my goal was to provide positive real-time news. With this update to Hope, we’ll dive into our hopeful past.

For example, did you know that the percentage of parliamentary seats held by women has almost doubled since 2000 alone?

In the updated version of the website, you will find the Hopeful Past tab. There you have three means of visualising our past: through an interactive timeline, decade cards, and my favorite: progress charts. You can explore events we can genuinely be proud of, going all the way back to 1800.

Because many visualisations only have high-quality data from around 1950, I added an uncertainty label to the charts. It shows from which year the data is verified, with a high probability of certainty. I used a range of public datasets for this project. Two of the main platforms were Our World in Data and the World Bank.

Some other useful features:

+ References and Further Readings per visualization

+ Category filter

+ Methodology and Download Data icons per chart

+ Chart Summary percentage on the top right of every chart

+ Click-and-drag to select shorter timespans, which recalculates the Chart Summary percentage

If you’re interested in this topic and want to learn more, here are a few favorites:

+ ▶️ Hans Rosling Ted Talks (links in the comment 👇 )

+ 📚Factfulness by Hans Rosling

+ 📚The Better Angles of our Nature, by Steven Pinker

Link to Hope: https://lnkd.in/enywdVh4

Link to my original LinkedIn post (introducing Hope 1.0): https://shorturl.at/zgKVg

Agentic Engineering Solution #4: Tracking the state of the AI bubble

While the irony is not fully lost on me, asking AI to help me create a website that explains the AI bubble was the best way to investigate it.

Among others, on bubble.arian-shafa.xyz you will see:

1. Financial information about 19 of the most prominent AI companies, throughout the stack: from the infrastructure layer (NVIDIA, ASML, TSMC, etc.) to the Model layer (OpenAI, xAI, Gemini, etc. ), to the Application layer (Cursor, Lovable, etc.). You can filter the entire page on one layer or all three.

2. This includes their valuation, revenue, profit, and funding rounds. All information is automatically refreshed.

3. One of my favorite sections is Claims Analysis, showing their historical claims categorized as successful or failed.

4. Throughout the website, clickable source icons link to fact-checking sources, and each visualisation includes a certainty tag indicating the trustworthiness of the research and sources.

5. And, there is a whole forecast section (incl. the methodology) which shows how absurd some of these numbers could look a few years from now (~1 trillion in revenue for NVIDIA by 2030?!).

Other notable features

1. The Your Dashboard section, where you can build the type of KPIs you’re most interested in.

2. The Comparison Mode, which allows you to compare the performance of up to three companies.

3. Most of the visualisations and tiles are downloadable, so you can play around with the data yourself.

4. The must-have dark/light theme toggle. ;)

How can I improve it?

Agentic Engineering Solution #5: Financial Literacy

I was lucky to stumble upon a classic financial literacy book when I was 16 (Rich Dad, Poor Dad; thanks Sunny Colson for introducing it to me on the metro). That, in turn, led me to be more intentional about money and long-term investing. This was one of the best things that could've happened to me financially.

Most people I talk to agree that we don’t teach enough financial literacy skills to our youth. This is why I created finlit.arian-shafa.xyz.

In it, you can find:

1. How small, good habits, when done consistently, compound over time (in the interactive Compound Interest Calculator; screenshots 3 & 4).

2. A Learning Hub that teaches five fundamentals of personal finance (budgeting, saving, investing, debt, and big life decisions).

3. A life simulation game where you can directly see the financial impacts of your choices.

4. A curated Resources section to learn more, which includes the resources I’ve personally found useful.

5. And my favorite feature: in the Learning Hub, each section allows you to personalise your learning and get strategies that are more tailored to your situation. All you need to do is click the “Personalize this Section” button (see the last screenshot).

What would you add to this website?