How Much AI Code Would You Use?

I saw an article recently that a quarter of the Y Combinator startups have 95% of the code in their repos being AI generated. The article notes that if no other startups had any AI generated code (no idea the likelihood here, then about 24% of their code for startups is GenAI written.

24%

Is that high or low? If you think about all the code you’ve written in the last year, how much of it could be reasonably generated by AI? All the queries, schema changes, test code, dummy data insertions, refactoring to add a column to a table or result set. Could it be AI written?

Let’s assume you practice a bit and you could quickly get some table, view, and sproc code. If I were to ask an AI to add a date column to a table and adjust the view and sproc code, is that useful? If I can move on to another task, or start setting up a prompt to add some test queries (or tsqlt tests) for me, that might make me more efficient. Certainly I’d still have to think of how to model this, but could the AI save time?

There is a lot of code that AIs might not write well. Certainly any dynamic SQL stuff that might be susceptible to SQL Injection worries me. An AI might write this, but so can a human. If the human doesn’t understand this issue, then it doesn’t matter if the AI writes it or the human does.

Except the human can make more mistakes in a period of time with an AI helper than by themselves. At least, I think they will unless they just can copy/paste/search/replace the code themselves. Then maybe they can do the same level of damage to a codebase.

The article has this great quote: “… even if product builders rely heavily on AI, one skill they would have to be good at is reading the code and finding bugs.” That is very true, and it needs to be said to every manager, project leader, and executive. You still need trained, skilled, experienced developers.

This might be even more true for DBAs and ops people using AI to help them write scripts to change infrastructure. Plenty of people do this poorly now, but how many more quick scripts written by an AI will include unintended, insecure, or otherwise problematic code? People need to use the AI as a tool and review the output before using it.

So how much help could you get from an AI in your coding? 24%? 95% Let us know today.

Steve Jones

Listen to the podcast at Libsyn, Spotify, or iTunes.

Note, podcasts are only available for a limited time online.

Posted in Editorial | Tagged | 2 Comments

Back in the UK

I flew overnight and am in the UK this week. It’s a time for me to get into the office and get some facetime with various other Redgaters while I’m over. I usually go a couple of times a year to the office and spend time with co-workers, which I find valuable to build bonds and relationships.

I was chatting with a friend recently who is remote. Most of his data team is remote, but there is a main office in another city. They are debating Return-To-Office guidelines, and they’ve required any new hires to be in the other city.

That’s a concern to this person, and others. Should he consider moving or look for a new job? Is there a lack of growth in this position if RTO becomes mandated. In this case, this person would rather have a remote team that gets together once a year.

Interesting idea, and I’ve seen this in the past when different positions when there were remote teams. It can be expensive, but it can be an investment. I’ll need to think about this and maybe write a piece.

Not much blogging this week, as it’s been a busy time, so I’ll be back next Monday with a set of more regular posts.

Posted in Blog | Tagged , | Comments Off on Back in the UK

Advice I Like: Tipping

When in doubt, overtip – from Excellent Advice for Living

This is close to my heart, since I spent a lot of time in college and after working as a waiter and bartender.

In the US we have a culture of 15-20% of tipping at restaurants. There are also guidelines for people handling your baggage, valets, etc. In the EU/Asia, often tipping isn’t a common thing, but that is changing a bit. I’m less familiar there, so this is my US view.

I like to tip for service, but I also know that many people in restaurants get a very low wage. When I started working as a waiter, I got US$2.01/hour. When I stopped, I was still getting $2.01/hr. In that time, the minimum wage went from $3.35 to $5.25. The reason my wage stayed low is that it was assumed I made tips that raised me to the minimum or higher.

My view is that if I can’t tip, I don’t go out. I tip 15% as the service for someone bringing me food and drinks. Even if the service isn’t great, the person deserves to be paid for their work and they aren’t really getting that from the establishment.

If they do a good job and are pleasant, I usually tip 20%. If they’re great I go higher.

However, if I’m not sure, or maybe I realize I might not be in a good mood or distracted, I overtip. Especially for breakfast service.

It’s a few dollars for me, but it makes a big difference to someone else’s life.

Note: The current post-pandemic world where every counter service place adds tips also annoys me. I think most of these people are being paid minimum wage and I don’t often tip, but I do when people are pleasant. Or I appreciate their being there. I always tip at the airport counters because those people are driving a long way and I really, really appreciate they are open at 6m for my early flights.

I’ve been posting New Words on Fridays from a book I was reading, however, a friend thought they were a little depressing. They should be as they are obscure sorrows. I like them because they make me think.

To counter-balance those, I’m adding in thoughts on advice, mostly from Kevin Kelley’s book. You can read all these posts under the advice tag.

Posted in Blog | Tagged , | 1 Comment

Republish: The Changing Nature of Data

Today I’m in Dallas, coaching 14 year olds in a volleyball tournament. This is our last travel trip with the team, and this is day 1. Hopefully we get a good draw and have some success.

While I’m wearing earplugs in a large convention center with 1,000 screaming teenage girls, you get to re-read The Changing Nature of Data.

Posted in Blog | Tagged | Comments Off on Republish: The Changing Nature of Data