Human vs. ChatGPT NFL Week 3 betting picks: can AI recover from its ugly 2-8 start to the season?
After going 0-5 in Week 1, GPT finally picked a game correctly. Actually, it picked two! But it still went under .500 and is now a dreadful 2-8 on the season.
I'm exactly .500 at 5-5, which isn't awesome, but looks fantastic next to GPT's putrid 20% win rate. It also makes me feel a lot better about backing the Steelers getting 3½ points against Cincinnati in Week 3, since AI is on the opposite side.
The rules remain the same. ChatGPT and I each make our five best bets using the same set of odds, and no selection can be priced worse than -120.
THREE WORD GAME: THREE WORDS TO DESCRIBE EVERY NFL TEAM SO FAR THIS SEASON AHEAD OF WEEK 3
Every wager is sized to win one unit, which means a bet at -110 risks 1.10 units to win one, while a bet at +110 risks 0.91 units to win one. Odds were captured Saturday, Sept. 26, at 6 p.m. ET.
Before we get to the Week 3 picks, let’s review last week.
The Browns rewarded our willingness to back them after an ugly opener by beating Tampa Bay outright, while the Panthers and Falcons comfortably stayed under the total with 37 combined points.
I'm going to whine about Jayden Daniels getting hurt, because that certainly didn't help our Commanders bet. But I was flat-out wrong about the Packers and Chargers. Green Bay was fortunate to win at all, and the Chargers made me look really dumb for expecting a bounce-back performance.
We leaned heavily on the idea that the betting market overreacted to Week 1. Week 2 reminded us that sometimes a team looks bad because it is bad. We learned from those mistakes, and we're ready to get back on track in Week 3.
The robot is officially on the board! After going 0-5 in Week 1, GPT improved by going 2-3 in Week 2. Both unders cashed easily, with the Steelers and Patriots combining for 23 points and the Vikings and Bears managing just 12.
The spread picks were another story. Jacksonville lost by seven, the Giants got blown out by 22 and Kansas City needed overtime to beat Indianapolis by three. GPT is now 0-7 on spread and moneyline picks through two weeks.
SEASON REPORT CARD
DAN: 5-5, -0.24 units
GPT: 2-8, -6.50 units
Tennessee Titans at New York Giants, Sunday at 1 p.m. ET
This is what's known as a "plug your nose" bet. It doesn't feel great to back the Giants after they just lost quarterback Jaxson Dart for the rest of the regular season and Jameis Winston looked atrocious in his stead. However, with a week preparing as the starter, I expect Winston to be better and the Giants to rally around their backup QB. Plus, I just don't think Tennessee is very good.
Getting the Giants below a field goal is the appeal, and Malik Nabers being cleared gives Jameis Winston his most important receiving option. My concern is trusting New York to create separation with a replacement quarterback on a short week. Brian Burns, Deonte Banks and Tyler Nubin are questionable, while Tennessee’s running game offers a way to keep this close. I passed because the Giants’ defensive uncertainty outweighs the attractive number.
Cincinnati Bengals at Pittsburgh Steelers, Sunday at 1 p.m. ET
This is definitely a "fade the public" type bet. The Steelers haven't looked good in either of their two games, despite being 1-1, and the Bengals look like legitimate Super Bowl contenders. So, why are we backing Pittsburgh? I love getting 3½ points at home in a divisional matchup. With everyone expecting Cincinnati to roll Pittsburgh, I'll grab the points.
The extra half-point is valuable in a matchup Pittsburgh’s defense could turn into a field-goal game. This is our direct head-to-head disagreement, though. I’m backing Cincinnati because its improved defensive front faces a struggling Steelers offense, while Joey Porter Jr. and Jamel Dean are questionable against Joe Burrow, Ja’Marr Chase and Tee Higgins. Pittsburgh can cover without winning, but I trust Cincinnati’s paths to creating a margin more.
Minnesota Vikings at Tampa Bay Buccaneers, Sunday at 4:05 p.m. ET
Honestly, the Vikings have been at their best with Kyler Murray not playing quarterback, and he's back to starting in Week 3. This just feels like the Vikings are a pretty fraudulent 2-0 and the Bucs are probably not quite as bad as their 0-2 record suggests. The lookahead line prior to Week 2 favored the Bucs, so I think we're getting a lot of value with Tampa as a home dog.
There is a reasonable case for looking past Minnesota’s 2-0 record and Tampa Bay’s 0-2 start. The Vikings have not consistently sustained offense, and Tampa Bay has left scoring opportunities unfinished. I also prefer the Buccaneers moneyline to taking just one point. What keeps me off the bet is Baker Mayfield’s protection matchup against Brian Flores’ pressure, plus the uncertainty of projecting Minnesota’s offense with Kyler Murray returning.
SIGN UP FOR THE DAILY OUTKICK NEWSLETTER
Las Vegas Raiders at New Orleans Saints, Sunday at 4:25 p.m. ET
This game has sneaky shootout potential, especially with the suddenly explosive Saints' offense led by Tyler Shough and Chris Olave. Entering Week 3, Shough led the NFL in passing yards, while Olave ranked second in receiving yards. The potential return of Brock Bowers would boost a Raiders' offense that has quietly produced 53 points through two weeks despite missing its best offensive weapon.
Tyler Shough and Chris Olave give this over a credible explosive-play argument, and a Brock Bowers return could raise Las Vegas’ offensive ceiling. Bowers remains questionable, however, and the Raiders have preferred long possessions while struggling to generate consistent yardage. Their pass rush also faces a Saints line missing left tackle Kelvin Banks Jr. I’m backing Las Vegas +3, but I’m less confident these offenses combine for at least 45 points.
Philadelphia Eagles at Chicago Bears, Monday at 8:15 p.m. ET
Following a historic Week 1 offensive performance, Chicago couldn't back it up in Week 2 and scored just 3 points against the Vikings. With Caleb Williams set to miss the game and backup quarterback Tyson Bagent also questionable, it's hard to see how the Bears score more than 17 points against Philadelphia. On the other hand, the Eagles' offense hasn't been very good, either. Finally, primetime unders have been profitable since the start of 2016.
This is the closest of your five picks to making my card. Caleb Williams is out, Chicago has not confirmed whether Tyson Bagent or Case Keenum will start, and Philadelphia is missing Dallas Goedert. Those absences support a lower-scoring game. My hesitation is that Chicago can still attack Philadelphia’s vulnerable run defense, while Saquon Barkley and DeVonta Smith are cleared. I lean under, but the quarterback downgrade is already reflected in the number.
For full transparency, I used GPT-6 Pro, powered by GPT-6 Astra, and set the thinking effort to Max. Then, I fed it the OddsChecker board as it appeared Saturday at 6 p.m. ET and gave it the following prompt:
"Pick your five best bets for Week 3 based on the odds provided. No bet can be priced worse than -120. Review every spread, total and moneyline that falls within those parameters and determine whether it belongs among your five picks. Explain exactly why you’re selecting each of the five bets. Also explain exactly why you’re passing on the eligible bets you don’t select. Use at least 10 different sources for each game so you have a broad understanding of everything surrounding each game. Before delivering the results, do a fact-check and ensure that everything is correct."
After nearly 30 minutes of thinking, here’s what it came up with.
Los Angeles Chargers at Buffalo Bills, Sunday at 1 p.m. ET
Buffalo has scored 77 points through two games, while the Chargers have managed 14 in each of theirs. The Bills also have additional preparation time after playing Thursday in Week 2. Los Angeles’ protection problems and confirmed absences on defense give Buffalo several ways to build a lead. DJ Moore and Keon Coleman being questionable is the main concern, but I still favor Buffalo at a number that pushes on a touchdown win.
GPT’s projected score: Bills 30, Chargers 20
This feels like a trap. Buffalo should wipe the floor with the hapless Chargers, but that's exactly why I wouldn't go anywhere near this spread. Bets that seem obvious usually aren't.
ROB GRONKOWSKI CALLS CHARGERS 'PROBABLY THE BIGGEST MESS IN THE NFL' RIGHT NOW AFTER ROUGH START
New England Patriots at Jacksonville Jaguars, Sunday at 1 p.m. ET
Jacksonville’s defensive front can exploit New England’s protection after Mike Onwenu’s move to injured reserve. The Patriots are also missing edge players Dre’Mont Jones and Quintayvious Hutchins, making it harder to pressure Trevor Lawrence without additional help. New England’s defense remains the strongest counterargument, particularly after Jacksonville struggled in Denver. At even money, though, I’m willing to back the Jaguars with a push on a three-point victory.
GPT’s projected score: Jaguars 24, Patriots 20
I like this look because one would think the public would love getting the Patriots as a dog in this spot. Betting against the public is generally good strategy.
Cincinnati Bengals at Pittsburgh Steelers, Sunday at 1 p.m. ET
Cincinnati’s rebuilt defense has eight sacks through two games and just held Houston without a touchdown. Pittsburgh comes off a three-point performance and has uncertainty at receiver, running back and both outside cornerback spots. Those questionable players could play, but Burrow, Chase and Higgins give Cincinnati the more dependable passing combination. Laying more than a field goal is the risk; the relatively inexpensive -104 price helps make it worthwhile.
GPT’s projected score: Bengals 24, Steelers 17
Well, I bet the Steelers, so you can probably guess that I don't like this pick.
Las Vegas Raiders at New Orleans Saints, Sunday at 4:25 p.m. ET
Las Vegas has eight sacks, and New Orleans has allowed eight while now losing left tackle Kelvin Banks Jr. That matchup gives the Raiders a way to disrupt Tyler Shough without needing their own offense to dominate. Shough’s explosive passing and Brock Bowers’ uncertain availability are reasons to take the points. At +102, a narrow Saints win can still cash the bet, and a three-point Raiders loss returns the stake.
GPT’s projected score: Saints 23, Raiders 21
I don't love this one, either. In a potential shootout, I don't love the 3-point underdog. If anything, I'd rather bet on the Raiders moneyline for the outcomes where they outscore the Saints. But I wouldn't touch this spread either way.
Los Angeles Rams at Denver Broncos, Sunday at 8:20 p.m. ET
THE CRAIG CARTON SHOW–UNFILTERED, UNAPOLOGETIC AND UNMISSABLE. DOWNLOAD HIS DAILY PODCAST NOW!
Denver’s defense presents a tougher matchup than the Giants did, and the Rams travel on a short week with Puka Nacua doubtful. The Broncos’ offense has also struggled to sustain drives, while Los Angeles has strong coverage personnel and Aaron Donald back. Davante Adams and Jaylen Waddle create explosive-play risk, and 44 offers less cushion than the earlier 45½. I still favor the under, with exactly 44 returning the stake.
GPT’s projected score: Rams 23, Broncos 20
This is probably the correct side. Primetime games tend to go under and people see the Rams and think "points." But the Denver defense is legitimate and Puka Nacua is likely out. Plus, the Broncos' offense is low-key below average.
OK, that's it for Week 3! Check back next week to see whether GPT can finally win more bets than it loses. I'll be here, hopefully with fewer excuses (but probably not).