Tuesday, December 5, 2023

Hard to Visualize Novel AI Use Cases and Firms

Thinking back to 1995, it seems obvious that people had a hard time imagining what would be possible and how business models, behavior and products could change as the internet took hold. 


Until 1994, for example, no visual web browser had become popular. Internet Explorer was not released until 1995. Prior to about 1993, the internet was text-based. Only with the commercialization of web browsers could sites develop multimedia capabilities. 


But multimedia experiences were limited by internet access bandwidth as well. Generally speaking, home broadband speeds did not become widespread until about 2001, using the traditional definition of 1.544 Mbps as the minimum for “broadband” connections. 


Year

Typical Speed (kbps)

1995

28.8

1996

33.6

1997

56

1998

128

1999

256

2000

512

2001

1024

2002

2048

2003

4096

2004

6144

2005

8192


Keep in mind that Amazon, for example, was founded in 1994, and only sold books. Netflix was founded in 1997, but only shipped compact disks. Google was not founded until 1998. Facebook did not emerge until 2004. 


You get the point: early on in any new era, it is hard to envision what will emerge. 


Company

Founding Date

Amazon

1994

Alphabet (Google)

1998

JD.com

1998

Meta (Facebook)

2004

Alibaba

1999

Tencent

1998

ByteDance

2012

Netflix

1997

Meituan

2010

PayPal

1998

Salesforce.com


1999



We can reasonably assume some new forms of infrastructure will be necessary, as internet service providers and cloud computing “as a service” were necessary for the broadband-supported and multimedia  internet. 


Harder to determine early on are the emergence of new applications and use cases, as was the case for search firms, e-commerce giants, social media, video and audio streaming services, ridesharing, software and computing “as a service,” fintech, the gig economy, cybersecurity and many forms of analytics are some of the new industries created as a result of the commercialization of the web, multimedia web and broadband-supported web


Obviously, artificial intelligence likewise will create some new industries and functions, many related to the “infrastructure” part of AI, as was the case for internet service providers, cloud computing “as a service” firms and mobile apps. 


Other trends might not create specific companies or industries but underpin all of them. At a broader level, the multimedia and “easy to use” nature of the internet was enabled by the World Wide Web, which was an enabler, not a “company.” The web, and the internet, in turn, was built on part on appliances, including mobile phones and personal computing, which did create specific companies. 


The point is that architectures often do not create companies, but infrastructure does. And devices and apps virtually always lead to company creation. All of which might seem pedestrian, but are quite relevant as we try and envision the changes AI will bring. 


Most firms and processes will change because of AI, but that does not necessarily create “new” industries or functions, except in infrastructure areas. All firms use computers and the internet and mobility, for example. 


But some new industries have arisen with personal computing, the internet and mobility, and likely will also happen with the emergence of AI. The trick is imagining what new industries and functions could arise, aside from AI reshaping all existing industries and functions.


No comments:

Have LLMs Hit an Improvement Wall, or Not?

Some might argue it is way too early to worry about a slowdown in large language model performance improvement rates . But some already voic...