Daily one-pagers about AI, leadership, technology, and being human.

johnmaconline

I'm writing to think, learn, and remember in public. I'll be here everyday.

August 19, 2024 1 minute read

Moravec’s Paradox

Moravec’s paradox is the observation that some tasks that are easy for humans are very difficult for AI systems to replicate, while some tasks that are challenging for humans are relatively easy for AI. 

We also call this the “easy-hard problem for AI.”

Here are some tasks that are (generally) easy for humans but hard for AI:

  • Recognizing objects in complex real-world scenes (finding the squirrel in the woods)
  • Understanding natural language in context (sarcasm, idioms, etc)
  • Transferring knowledge to new tasks (learning the guitar after learning the piano)
  • Reasoning about abstract concepts (justice, empathy, etc)
  • Common sense understanding (if you have an umbrella, I know it’s probably raining)

Here are some tasks that are easy for AI but (generally) hard for humans:

  • Strategy games (chess, go, etc)
  • Complex calculations (weather forecasting)
  • Large datasets (genomic data)
  • Predictive modeling (protein folding)
  • Image recognition (MRI analysis)

The good news for us humans is that we can use AI to augment our capabilities for the hard (for us) tasks. You can and should be doing that already. 

Will AI someday use humans to augment its capabilities for the hard (for it) tasks?

August 18, 2024 1 minute read

Reasonable Decisions and AI

Better safe than sorry.

But…

The smoke detector shouldn’t go off because we burned dinner.
The car shouldn’t automatically stop in the middle of the road because of a pedestrian on the shoulder.
The security system shouldn’t call the Police if the dog trips the alarm.

These decisions aren’t simply a math problem for humans.

We humans are good at the nuance of detection. We make an assessment of the situation with all of our faculties. If we’re in charge of the decisions above, we know how to quickly and accurately process all of the signals provided to ensure we make a reasonable decision. Reasonable does not have to mean “better safe than sorry.”

AI’s not (yet) good at processing the complexity of real-world systems. Even Tesla’s full self-driving (which is amazing — I’ve used and witnessed it myself) is much more conservative than the most conservative driver you know.

AI can’t think, or at least not the way we do. AI doesn’t care.

That’s the difference. We humans can make reasonable decisions because we care.

Don’t stop caring.

August 17, 2024 1 minute read

But Does it Make a Difference?

Does it make a difference if you blow the red light when nobody is around?
Would it have made a difference if the US boycotted the 1936 Berlin Olympics?
Will it make a difference if I have another beer?
Does it make a difference if we contract out this piece?
Would it have made a difference if I had made the bed?
Will it make a difference if we elect (whomever)?

Sometimes, leaders get caught up trying to answer the question of whether it makes a difference. Here’s the Cliffs Notes: the answer is always “yes.”

The question you should be trying to solve is, “To whom does this make a difference?”

August 16, 2024 1 minute read

The Pinocchio Problem and AI

Pinocchio says, “My nose will grow now.”

But Pinocchio’s nose only grows if he’s lying. Therefore, if his nose grows, he’s lying. Conversely, if it doesn’t, he was lying, thus making the statement true again.

The AI says, “I always tell the truth.”

But who’s truth? And when?

August 15, 2024 2 minute read

The Class Action Incentive Problem

I just signed up as a claimant in (yet another) class action lawsuit. This one is against Visa and MasterCard.

“The deadline for the Visa/Mastercard $5.54 billion settlement claim is approaching quickly…This settlement provides all eligible businesses the opportunity to reclaim a portion of the interchange fees paid between 2004 and 2019.”

Class action lawsuits, like many other boondoggles, are good-intentioned at their core.

The primary purpose of a class action lawsuit is to allow a group of individuals to band together and sue collectively against larger entities. If Big Evil Company has misrepresented, mischarged, or otherwise duped us little people out of what we deserve, then we can get together and stick it to the man.  

Power to the people, and whatnot.

Y’all know how it works in practice. The law firm gets a pile, and we, the claimants, get a tiny drop. 

“Hey! That’s not fair!” 

Sure it is. It’s actually the very definition of fair.

I don’t fault the law firm, nor am I bothered by the distribution. If I was duped out of $3.64, then I should be repaid $3.64. The law firm does the work on behalf of all of us. They should get paid. 

The problem is the incentive. 

It’s the Cobra Effect, the social media algorithms, and KPIs

We usually get what the system incentivizes.  

August 14, 2024 1 minute read

Pointing at Who Did It

The pilot who successfully lands the crippled airliner gets the headlines. So does the emergency room doctor who restarts the heart of a 53-year old man.

As they should. These are extraordinary efforts that lead directly to saved lives.

Plus, it’s easy to point at “who did it.” Headlines love who did it.

What about the engineer who designed a redundant switching mechanism in the aileron control system? That system has never once failed over the lifetime of the aircraft fleet on which it has been installed for over 30 years.

What about the 53-year-old man who changed his lifestyle ten years ago and is now healthier than he was at 40?

If nothing bad or dramatic happens, it’s harder to point at who did it.

But that doesn’t mean somebody didn’t do it. As a leader, your job is to look under the covers, past the mundane, and point at who did it.

August 13, 2024 1 minute read

Benevolent Monopoly

The benevolent monopoly can make life easier and better for customers. That’s why they’re benevolent.

Bell Telephone, Standard Oil, and Microsoft all were benevolent monopolies. The efficiencies of their monopolies were passed on to customers in the form of lower prices, standardized practices, and consistency in quality.

Walmart and Costco are not technically monopolies, but they kinda are, and they’re benevolent. If you’re a widget or foodstuff (and I use foodstuff in the very loosest way possible) producer, you want to be on their shelves. They are also benevolent. Consumers love both of them, mostly because of price.

But is it really better?

What innovations never see the market because they decide? What supplier doesn’t get a chance to make a difference because they can’t get in the door? What ideas never see the light of day because they don’t fit their vision?

The problem with benevolet monopolies is the same as with plain-ole monopolies.

August 12, 2024 1 minute read

Do You Want to be Boeing or SpaceX?

Boeing has pedigree. SpaceX is the young upstart.

Boeing is slow and steady — do it right the first time. SpaceX fails fast and fixes it later.

Boeing conjures feelings of pride, trust, and safety. SpaceX’s CEO pisses off the public regularly.

Boeing is inclusive. SpaceX is divisive.

Currently, Boeing has stranded astronauts at the space station. SpaceX may be the one to rescue them.

Who do you want to be?

August 11, 2024 1 minute read

Are You Committed or Just F’n Around?

James Hetfield said that when he found Lars, he had finally found the other person who was just as committed as he was to making it work. Everybody else was just f’n around.

That’s who he was looking for. Someone else who was committed. He knew this was who he was going to hitch his wagon to, regardless of their differences.

You know the difference. You see it in others. If you’re honest with yourself, you see it in you. You know when you’re committed or just f’n around.

Commitment overrides differences. It engages complementary skill sets. It makes 1 + 1 > 2. It breeds trust. It pulls through.

Just f’n around is fine. We can’t be committed to everything in our lives all the time. Nothing wrong with that, but don’t kid yourself.

Success, whatever that looks like, requires committment — marriage, startups, bands, hockey teams, congregations, communities.

Are you committed or just f’n around?

August 10, 2024 2 minute read

The Future of Software Engineering with AI

More. Better. Faster. Cheaper.

  1. Every software engineer will have an army of coding interns who rarely get it wrong. 
  2. Therefore, every software engineer will be an architect, chief engineer, and engineering manager. 
  3. You may think this means less software engineers, but it will be the opposite. Today, software is expensive and slow to create. Unshackling army size and cost will lead to an exponential increase in software applications. More software engineers will be required.
  4. Given the changing software engineer’s role, so will his/her language (more natural language, less syntax) and skillset (vision, requirements, higher level problem thought, etc).
  5. The coding interns (ie, the AI generative coding models) have a great advantage over other generative AI use cases — namely, they live in a closed system with specific purpose and requirements. Given an endpoint and purpose, AI can easily write code, generate tests according to the purpose and implementation, execute tests, and then adjust the code until the requirements are met. 
  6. Product quality will increase, but probably after some famous and widely reported bumps in the road. 
  7. The coding interns bring another advantage — they can quickly try adjacent use cases and what-ifs, which helps the software engineer rethink purpose and outcomes. 
  8. The coding interns will get good at making decisions about applications and services to use to solve problems. Will we start marketing to AI? I think so.
  9. Similar with UI. AI uses the CLI (just like me) and API. These will become even more important than today. 
  10. Startups, especially solopreneur startups, will increase. New ideas, applications, and creative endeavors will become more accessible to more people through software abundance. Business will follow. This might be, at the end of the day, the most important and largest effect.

Pin It on Pinterest