We asked the AI robots to build a March Madness bracket for us and the best one was made by a newspaper

The process seemed simple enough.

Over the past few weeks we've seen AIs used to build blog posts, respond to Hinge matches, and gaslight journalists for being a "bad user."

Surely an artificial intelligence as manipulative as this could see through the chaos and find me a winning bracket.

Since ChatGPT's data set ends in 2021, rendering it useless to predict results for 2023, we turned to Microsoft's now-famous chatbot.

Alas, the results were a struggle.

Bing wouldn't fill out a full bracket for us, but it did provide some guesses at a Final Four.

While we had hoped for a full bracket, our first request to the AI only got one pick out of each region as a response.

"That's a tough question to answer because there are so many factors that can affect the outcome of the tournament," the bot told us. "However, based on some of the predictions I found online, here are some possible contenders for the best men's March Madness bracket this year."

Here is Bing AI's Final Four: No. 1 Alabama out of the South region, No. 1 Houston out of the Midwest region, No. 1 Purdue out of the East region, and No. 3 Gonzaga out of the West region.

Its reasoning was generic, but not incorrect. "Gonzaga is a popular choice as they are one of the few teams that have never won a national title but have been consistently dominant in recent years," Bing said. "They have a balanced offense led by Drew Timme and a favorable draw in the West region."

All true.

But I am here to make money, and barring going four-for-four on these Final Four picks, it's tough to get rich going all chalk.

Many of the sites that host bracket challenges — ESPN, CBS Sports, etc — have autofill options for users that are out of the loop.

Above you will see the options available to bracket builders on ESPN. Going fully chalk is boring, and going totally random results in a bracket doomed to fail before it can even begin, but the latter two options — random weighted and BPI — both offer some appeal.

Random weighted means that the ESPN machine is rolling some dice in the back of its mind and is aware of the fact that it's not uncommon for a double-digit seed to reach the Sweet 16.

Smart Bracket powered by BPI offers some of the same wonder that the potential of an AI bracket previously held. I don't know how ESPN calculates their BPI, but it has to be smart — it says so right there in the name!

We tried both.

This is an enjoyable bracket!

An Alabama-Houston final is a bit chalky, as they are the top two teams in the tournament, but UCLA making it out of their region is a bit fun.

The real chaos evolves in this bracket's East region, where No. 8 Memphis and No. 11 Providence will meet in with a spot in the Final Four on the line. Nice! Bold, but nice!

We've seen such wild matchups before. Last year, the East region Elite Eight put No. 8 North Carolina up against No. 15 St. Peter's. Sometimes weird things happen!

Of all the options the computer spit out, this was the best.

Again, the inner-workings of ESPN's BPI metric remain a mystery, but the first bracket it spit out was usable, if not a favorite.

There are some good upsets here, and a Final Four of No. 1 Alabama, No. 6 Iowa State, No. 4 Tennessee, and No. 2 UCLA has some appeal, satisfying the chaos that often comes with March Madness.

But this feels like a bracket that winds up dead halfway through the first weekend of the tournament. Boldness is fun, but if you're buying into a bracket pool, you want to feel like you have a shot at experiencing at least two weekends of enjoyment.

Yahoo Sports hosts many bracket pools and is a solid host of fantasy sports in general.

But their bracket autofill options leave something to be desired. Who wants to autofill with smart schools? The Ivy League has won eight games all-time at the NCAA Tournament.

And party schools? What arbitrary metric is being used to calculate that?

We had to investigate.

A Final Four of No. 15 Princeton, No. 15 Colgate, No. 5 Duke, and No. 7 Northwestern.

If it hits, it would look genius. But this is just a total bust.

With a Final Four of No. 9 West Virginia, No. 8 Iowa, No. 9 Illinois, and No. 7 Michigan State, this is another bold bracket.

Congratulations to Iowa for apparently being the best partiers in the country, but a winning bracket this is not.

Don't do this!

CBS Sports has an incredible SEO team, and always winds up extremely high on the search results for anyone Googling "who will win team x vs team y." It's good business!

But those links usually include invitations to join SportsLine, a betting information service that requires a bit of cash. Casual bracket builders have no use for such technology.

Similar to ESPN, CBS offers chalk options, as well as a historical random bracket, which accounts for the existence of upsets.

CBS also offers the opportunity to autofill with the brackets of several writers across their site, which is a fine option for someone looking to throw up their hands and just get something simple.

Our historical random bracket was a bit too safe.

While putting No. 13 Iona through to the Sweet 16 is a fun choice, as is putting No. 7 Texas A&M to the Elite Eight, the bracket ultimately played it safe, with all four 1-seeds reaching the Final Four.

All four top seeds have reached the semifinals just once in the modern history of the tournament.

FiveThirtyEight's sports coverage is fun to play around with.

Their bracket tool allows you to fill in potential matchups and see the percentages on any possible game in the tournament. If you have a half-filled bracket and are just wary of a few picks down the line, it's a fantastic resource.

However, using it to build the entirety of your bracket is less fun.

FiveThirtyEight's algorithm has a solid, but chalky Final Four, featuring No. 1 Alabama, No. 1 Houston, No. 1 Purdue, and No. 3 Gonzaga. It also features a Sweet 16 with no double-digit seeded teams reaching the second weekend of the tournament.

A March Madness without a Cinderella is no Madness at all.

Again, it's a great tool for double-checking your bracket before submitting it, or filling out areas where you don't have a preferred guess, but you'll want to build out at least half of your bracket on your own before asking Nate Silver for advice.

The Wall Street Journal's bracket builder offers just the right amount of user input to make your bracket feel like your own.

First, users decide between elevating the importance of offense, defense, or a balanced attack. Then they are allocated 60 points to distribute across five factors of play — size and rebounding, three-point shooting, turnovers, free throws, and tournament experience.

This allows casual bracket builders to make relatively informed choices without being overly informed.

The Madness meter.

How chaotic a bracket would you like to see? Again, this offers users the chance to get something that feels more personal. You can push the lever for a fully chaotic bracket, or play conservatively, but the choice is yours, not Bing's or BPI's or any other algorithm.

So much power in one little slider.

WSJ also gives you the chance to push the odds of one team in your favor, ensuring that if your alma mater is playing in the tournament, or if you have a pick to win but not much else, that team is properly accounted for in your bracket.

Just enough chaos. Just enough boldness. Apparently that Madness meter was in just the right place.

Here we have a Final Four of No. 8 Maryland, No. 3 Xavier, No. 3 Gonzaga, and No. 3 Kansas State. That's a pretty out-there Final Four, but it's not impossible. Further, it feels like the type of Final Four that would keep you alive in your bracket pool even if you only hit two of the four teams.

There are upsets in the early rounds too — No. 12 Oral Roberts upsets Duke in the first round and makes it all the way to the Elite Eight, and No. 13 Iona reaches the Sweet 16.

While one might want to dial things back a bit to send either No. 1 Houston or No. 1 Alabama a few games further into the tournament, this is a solid bracket, and at the very least, the robot/algorithm most worth going back to for another roll of the dice.

Just make a bracket! We don't need computers!

As much as computers already do for us, and as much as artificial intelligence is threatening to do in the near future, we don't need them to build our brackets for us.

If you want a resource, FiveThirtyEight is a nice way to sharpen your bracket, and the Wall Street Journal offers enough user input to feel like you had a say in the matter, but ultimately, it's all randomness.

Your guess is as good as any computer.