WEIZENBOCK CHALLENGE! AI vs. Grand Master Brewer

[@TheApartmentBrewer]

Thats a big starter for a big beer and may have contributed to a lack of expected esters, but stressing the yeast for the esters is playing with fire. Would be interested to see the impact of open fermentation! Still, sounds like a blast to brew! Matthew is an awesome guy and I would fully trust his recipe as well.

[@MeanBrews]

They were both good beers but i stand by my blind tasting. To me the ai beer had an almost worty taste to it. The meanbrews beer definitely had a more complex malt profile which i attribute to the addition of special b which ai missed. Ester formation is really dependent on pitch rate and and temperature. If youd like more banana then i would ramp the temps earlier. I actually got more esters on the aroma from the mb beer but flavor they were similar ester profile. The biggest reason for the starter is to keep from having bottle bombs during aging. I underpitched two batches and had to dump them 6 months in as they became bottle bombs

[@FLOYDMSMC]

Just curious as what level of carbonation you attempted the bottles at?

[@MeanBrews]

@@FLOYDMSMC3.1

[@TheBrulosophyShow]

In my non-blind tasting I came down on the side of the Mean Brews beer too. Let’s try again next year with a newer version of AI 😃

[@MeanBrews]

I'm up for it!

[@samstock6565]

@@MeanBrewsyeah that was fun to sit through esp when I’ve one in the fermenter as we this came out 👍

[@alex10905]

Great video. I really appreciated watching the judging. Makes sense what they were saying about the weizen yeast characteristics were muted in Matt’s since you had a 5 day starter.

[@MeanBrews]

The Meanbrews recipe just snagged a gold medal at the Dog Dayzz competition in Kansas City brewed by Mike and Stephanie Butler. They scored a 44/50 and did use a starter. Would be curious how the chat gpt beer would perform in competition!

[@TheBrulosophyShow]

That’s fantastic!

[@WarfareJournal]

Definitely need to enter an AI recipe into a competition and see what happens!

[@JohnL9013]

This was a fascinating comparison. Thanks for including all the details. I hope to see more collabs with Matt!

[@AlbrechtL]

I have seen MANY of your videos, and this one is probably my favorite so far. The format of the video was more entertaining for a variety of reasons and was engaging to get a range of opinions from certified judges (as opposed to just your friends). Their descriptions are far more nuanced and descriptive. It would have been interesting to have the judges provide their scores as well!

[@tobarstep]

The thing with AI/LLMs is that they are mostly just amalgamation engines. It did pretty much what Mean Brews did and just take a sampling of all the recipes it could find and "average" them out.

[@FLOYDMSMC]

first thing i thought when seeing his recipe was the same as what the judges said. my experience with hefeweizens is to underpitch it to get those desired falvours

[@wedomusic9451]

I have heard that too about hefes... the more stress the better for the style.

[@ElementaryBrewingCo]

Great vid!!! Sounds like I need to brew 1, or both, 😂 cheers 🍻

[@DimpieDeBruyn]

That AI beer looks way better. ... I will try brew that one sometime.

[@tomwidauer8376]

I've developed a recipe with ChatGPT. I gave it the initial parameters like what I wanted and which hops to use and then let it do its thing. I requested a few changes here and there and when it gave me what looked like what I wanted, I adjusted it a little to suit my brewing system and tweaked things a little more to suit my palate but the beer was great. I named it Skynet Pale Ale to honour the AI creating it. It's a shame I gave it to my father-in-law as he had just run out of beer and my kegerator and the keezer were full. But I had a few over time and it did mature well. I will brew it again, not as a priority as we are moving into winter down here and it clearly was a summer beer but it was good enough to brew again. I'm tempted to collaborate with ChatGPT a little more for different styles.

[@jac540]

A very general question: in a lot of your video's I see that during the mashing the outlet of your pump gives a very spray like profile. Does this gives a risk of hot side oxidation? The brewing kettles we can buy in the Netherland tend to have nozzle that concentrates the wort in one laminar flow, and a filterplate on top of the mash takes care of spreading the wort over the malt bed. What is the benefits and downsides of having the spray nozzle above the laminar flow nozzle with a filter plate?

[@PatrickBuckles]

Great vid. Makes me wonder if the consensus together would have been the same individually.

[@eltorpedo67]

great point. I've never been to a beer judging - I was wondering if they always confer amongst each other like that before scoring. Seems a bit weird to me.

[@PatrickBuckles]

@@eltorpedo67 ultimately they shouldn't be. I have seen a wide variety in scoring while some are pretty dead on with eachother. I've never seen the actual judging, but either its not consistent or the wide scale means they didnt discuss. Now you have me even more curious and I have to find out

[@Kaesekuchen002]

I really enjoyed this 2 video format and would love to see more of these examples. Maybe source the "AI recipe" from different sources. Might as well RNG generate some of the percentages or IBUs.

[@HonicBlue]

fun video. It also goes to show you that you can overcomplicate a beer recipe to the point of negative affects. More isn't always better, and to be honest, I steer clear of any recipes that get overly complicated.

[@goodolarchie]

Hey Martin this was a fun experiment. I am curious about the judges comments on yeast expression and the fermentation itself. In my experience the best hefe/dunkelweizen or weizenbock recipe won't overcome a meager fermentation, that well balanced isoamyl acetate and 4vg (banana and clove respectively) comes from both a ferulic acid rest, slightly underpitching the yeast, and keeping the temp in the mid 60's until after high krausen. Or at the very least - certainly not building a starter for it lol. I guess my take away is the recipe (i.e. grist and hops) isn't really the primary dependent variable here, fermentation was. I'd be curious if it were repeated but used identical fermentation parameters!

[@omarpadilla4739]

I never put o2 in my weizens. I spin a low gravity starter (1.018 ish) 12 hrs prior to pitch time. Tons of banana despite 5 psi press ferm. WLP300 yeast. Clove forward requires lower ferm temps (hold closer to 62) vs banana forward (66 and up). And obviously cant make clove without ferulic acid as well.

[@icemanknobby]

Watching your videos when youy doing a mash you use a balloon whisk, can you give me an idea of the length of that whisk? Have been trying to locate one but cannot decide if the one I have my eyes on is big enough:-)

[@enbi2499]

i brew with chat gpt a lot and i have to say one thing is very important. you have to bring in your own experience and have a conversation about what you want to do. sometimes i spot a tiny mistake that chat gpt would make. not really a mistake but the funny thing is, when you ask him "wouldn't it be better" like this and that he would then tell you yes you are correct because of this and that and knows even what my intention was. so if you bring your own experience and thoughts to chat gpt and you tell him very precisely what you actually want to make and discuss this recipe a lot then it becomes excellent.

[@paulbehrens5842]

I will have to give those recipes a crack!

[@sw1sha442]

the next logical step is to just vary the yeast starter/no-starter into one of the mash bills and see how much of a difference that makes

[@connorhulegaard2012]

I think the comparison isn’t great because like you said the chill haze was your issue. I think if you dealt with the chill haze you could make a better comparison of diptubes. Cheers.

Mayby can you try with Bard AI Google?

[@BurnedJello]

Now to make an AIPA

[@takger]

I didn't understand a word from 13:15 lol... but I'm an AI tutor, not a beer fermenter

[@KianaMassey-p8k]

Mayby can you try with Bard AI Google?. The Norm has spoken ....

[@mustyditch4703]

The Norm has spoken ...

[@sw1sha442]

it looked like you were getting zero mechanical advantage from your pulley arrangement when pulling out the mash basket. you need more than one cord in tension going down to the hook to generate mechanical advantage, that other pulley isn't doing anything.

[@flagadabilal6565]

Just one thing about pronunciation: in German z is pronounced like the ts in cats, not like a vocal s.

[@zsoltoravecz1815]

I think chat gtp is just learnt Recipes from the internet. Best way for ai beer, would be ro training an ai from many beer contests.

[@dearseall]

Weizenbock is pronounced "Weitsenbock" not Weisenbock ;)

[@N3ss3s]

I hope this is one of the last AI vs Human videos, it was a fun novelty at first, but all the AI is doing is scraping the web for bits of text and possibly just stealing people's work. Standing on the shoulders of giants and all that.

[@johnsikking4891]

But that is the same approach as the human beer recipe - he takes award winning beers are takes the statistical averages. ChatGPT is doing something similar.

[@N3ss3s]

@@johnsikking4891 Is it though? Nonetheless, my point stands, the entertainment factor of LLM is close to watching paint dry.

[@Duglum666]

I mean.. i get that you are not native speakers and your pronounciation won't be perfect.. but.. it's WeiZenbock! Not WeiSenbock! The latter just sounds completely wrong. Try to pronounce it like the Z on Led Zeppelin.. it won't be perfect but much closer.

[@stephenk1004]

Weitzen would be even closer again.

[@carlwaits1478]

"Grand Master" - failed to mention the accreditation comes from the program he started and manages with the master homebrewer program and cir-cult of America competition. Puts a different spin on the validity of his recipe.

[@TeeCee_DF]

How does his title affect the flavor of the beer? 🤔

[@jdauria61]

Yes he started it, but there is a Board of Directors that manages it now. But what's that got to do with his recipes? He does not even mention the program or his rank in Mean Brews recipes, Martin is the one that did. He just takes a bunch of award winning recipes, that he gets from numerous sources, crunches the data and comes up with a mean average recipe of them. Have used many of the recipes, they are great beers and have won plenty of medals with them. The New Zealand Pils one, in 6 comps I have scored 45,39,39,39,40,37...would like like to have more 40+ in that set, sure...but judging is subjective.

[@TheBrulosophyShow]

Matthew has consistently won a bunch of awards with his recipes and I’m lucky he worked with me on this.

[@MattMcHair]

Would love to know FG & ABV on these beers for consideration of how they came across, I don't think I saw/heard this info in this video or the @meanbrews beer. Maybe it's there and I missed it!