Generative AI is Changing the World. Now What?

My Own Thoughts on Artificial Intelligence from 2023 and Beyond

5 min readMay 6, 2023

Since ChatGPT emerged in November 2022 and entered the zeitgeist in 2023, I’ve been in awe of the developments within this space, and the massive implications it will have on all aspects of human life. Since 2018 (in my NS days), I’ve been reading up on AI, learning techniques and GPTs. But I’m convinced 2023 will be the year when GPTs and CNNs facilitate a turning point in humanity.

Why Now?

AI and GPTs have existed for a while. The former was invented over 50 years ago, while the latter was created about 5–6 years ago (in 2018). AI (and all its technologies) have been around for a long time. But I believe that today, as of Summer 2023, we are on the cusp of a seismic shift in how AI impacts our ways of life. This year will be the turning point in how humans live, because it is the year that AI penetrates into our collective ways of living at scale.

Why now? I believe 3 main drivers propel this shift:

Technological Capability
Accessibility and Product-Market Fit
Scalability

1. Technological Capability

Technology will continue to advance and be cheaper (in terms of unit economics), thus driving down the cost of innovation, computing and storage.

Information continues to be stored on scalable and distributed cloud-based solutions.
Computing and Processing will only be faster and cheaper.
Thus models with larger parameters can continue to be trained and invented.

In the past, technological capability was a key blocker, on the computing and training fronts.

AI development entered AI winter in the mid-1970s due to technological constraints.
Multi-layer neural networks sped up how models and machines “learnt”.
LLMs used to be trained via RNNs and CNNs, but the shift to Transformers exponentially grew GPTs’ capabilities.

Today these computing and training barriers are no longer as significant, and will only continue to improve from here.

While AI has existed for a long time (think TikTok and Amazon’s recommendation engines, computer vision and machine learning models), I believe the generative capabilities from CNNs has truly moved the needle.

2. Accessibility and Product-Market Fit (PMF)

This is akin to the moment when personal computers leave the laboratory and into the consumer market.

When launching a new product, it’s critical to ensure that the product lands and is adopted by the market. Accessibility and product-market fit are critical ingredients that contribute to a product’s long-term adoption and success.

ChatGPT reached 1 million users in 5 days, 100 million MAUs in January (2 months), and 1 billion visits in 5 months (March 2023). (Source)
In comparison, TikTok took 9 months and Instagram 2.5 years (Source) to reach 100 million MAUs.
Note: While I can’t find data on recurring users, I’m convinced it is a significant % of visits and MAUs.

These numbers are mind-boggling. Not just because of the unprecedented pace of growth, but because of its implications.

Launching ChatGPT is akin to releasing computers from the hands of secretive laboratories and corporations into that of ordinary consumers in the mass market. It’s equivalent to what Bill Gates and Steve Jobs did with personal computing.
The incredible pace of growth affirms that it has product-market fit among consumers. But data on recurring users, paid users and top demographics and segments will be even more helpful in validating this hypothesis.
These data and patterns reflect that mass-market consumers are aware, receptive and supportive of AI and GPT technology into their lives.

3. Scalability

Scalability via enterprise players and open-source community seals AI’s global adoption.

With mass market consumers aware and receptive to AI and GPTs, these technologies have secured a foothold among individual consumers in the market. But I believe the real catalyst of these technologies’ growth and global dominance lies in the players who are scaling these technologies.

As of May 2023, these are the biggest developments in AI news.

OpenAI enabled ChatGPT plugins, setting up an “App Store” and ecosystem of ChatGPT-integrated tools.
Microsoft invested into ChatGPT, and integrated GPT Technologies into Office365
Meta’s LLaMA model had been leaked to the public, enabling developers to build applications atop the model.
A Google researcher’s beliefs that OpenAI and Google will be outcompeted by open-source communities

These developments carry massive implications.

Productising GPT technology and building an ecosystem reduces barriers for corporations to integrate GPT tech into their processes, while weaving its adoption into humans’ ways of working.
Leaked models means innovation is no longer limited by corporations’ pace of development and internal resource capabilities/goals. This democratisation empowers open-source developers to accelerate use cases, build new tools and identify potential bugs faster than companies.
Open-source creators and developers will likely outpace development by large corporations, and remove their competitive moat (at least in the short-term).

Lastly — remember that these musings revolve only around GPT technologies. I haven’t explored the implications on other technologies like diffusion models and Stable Diffusion (although the 3 drivers above apply to Text-to-Image technologies too).

In sum, I believe these 3 drivers will continue propelling AI and GPT’s growth, creating a flywheel effect that will create a seismic shift in how we humans live and create value.

The best part? I believe we’re still extremely early in terms of realising just how GPT technologies can evolve. In terms of its product lifecycle, we’re likely still at the Introduction and Growth stages.

Counterarguments

While the growth drivers above illustrate how bullish I am about GPT technologies, I believe it’s critical to consider counterarguments and perspectives against this issue. Below are some counterarguments I can think of.

Who said all consumers are receptive? Many people fear AI and many are unhappy about its blatant plagiarism and infringement of other artists’ work.
Is GPT a fad like Crypto?
Won’t regulation hinder GPT’s development and adoption?
Won’t the cost of maintaining ChatGPT and innovating GPT technology stifle OpenAI’s operations?
AI Agents appear to just be building recurring iterations atop GPT technologies. Is this development really revolutionary/life-changing, or just hyped up more than it should be?

Caveats

We don’t know anything. Just a few years ago, the narrative on robots taking over our jobs was that physical/blue-collar work will be replaced first. Now it’s evidently not the case.