πŸ’‘ Musings of the Angry Optimist: Untrue established truths

πŸ’‘ Musings of the Angry Optimist: Untrue established truths

The discussion is full of pessimistic knee-jerk reactions regarding AI.

Mathias Sundin
Mathias Sundin

Share this story!

My thoughts, tips, and other tidbits that I believe are suited for a fact-based optimist. This newsletter is for you who are a Premium Supporter at Warp News. Feel free to share it with friends and acquaintances.

😩 Pessimistic knee-jerk reactions

Having thought a lot about pessimism, I now see three stereotypical knee-jerk reactions regarding AI. WALL-Y recently wrote a news piece about AI consuming much less energy than what's portrayed in the debate.

πŸ‘©β€πŸ’» ChatGPT uses less energy than an hour of Netflix
One hour of Netflix watching equals the energy required for between 300 and 3000 ChatGPT prompts. Ending your shower one second earlier saves enough water to compensate for 40 ChatGPT prompts.

Someone questioned this because it went against the "established truth," as he put it. The funny thing was that what he considered the established truth was based on the same sources as the figures we wrote about.

His established truth was actually a pessimistic narrative. He was right that it was established, but it wasn't true. (I must unfortunately admit that I myself partly fell for this narrative. I'll return to this in a column.)

This has happened repeatedly throughout history. Just remember the millennium bug that was supposed to crash IT systems when the clock struck midnight on New Year's Eve 1999.

Another pessimistic knee-jerk reactions is finding a flaw in new technology that humans commit to at least the same extent.

For example, when a self-driving car hits someone, it becomes huge news, but the fact that 1.2 million people are killed by other humans each year is just statistics.

The news about AI's energy consumption was questioned by another person based on it being written by an AI bot. These can hallucinate, which she's right about. The problem was just that she herself managed to stack no fewer than four factual errors in her own comment. Almost one error per sentence.

A third pessimistic knee-jerk reactions consists of getting stuck on an early problem with new technology and being unable to understand that it can be solved or become considerably smaller.

That ChatGPT and other large language models can make things up – hallucinate – is a real problem. But it's a problem that has become considerably smaller in the past two years and that we'll likely find a solution to, or at least achieve a level where AI hallucinates significantly less than humans.

Several people don't believe this. Moreover, they magnify the problem and claim it's a decisive flaw that means generative AI will never have a major breakthrough.

We've seen this kind of thing repeatedly in history too. Ethernet inventor Robert Metcalfe wrote in the mid-90s that the internet would "soon explode like a supernova and collapse." Electric cars are too expensive and who wants to buy a car that can only drive 200 kilometers before charging? In the 1800s, the first carbon filament lamps burned out after a few hours, which led contemporary newspapers to dismiss electric light.

Of course, you can't know that all problems will be solved in the near term, but when something is important to us, we're phenomenal at problem-solving. To assume the opposite is not smart.

These pessimistic knee-jerk reactions make it harder to understand our present time and even more so our future.

Mathias Sundin
Angry Optimist

❗ Other stuff

πŸš— Self-driving cars take over San Francisco

The green line is Uber and others, the blue line is self-driving taxis.

FrΓ₯n Mary Meeker's AI report.

🀺 For those of you who are old enough

β›½ Lead in gasoline

Lead in gasoline 1986.

Lead in gasoline 2021.