Actor / Filmmaker / Golfer / Anti-AI advocate #wolfman #godofwar #callofduty #apexlegends #hades #diablo #starwars
Joined April 2008
- Tweets 2,082
- Following 1,575
- Followers 26,621
- Likes 8,254
287 Photos and videos
Jun 4
I haven’t seen a burn like this since Muspelheim.
ok thanks for the feedback, dominos pizza corporate account
39
102
3,784
101,506
May 22
Gorgeous result here from the SAROS soundtrack. Take a listen. 🎧
May 22
Recording a soundtrack inside a stone church!
Join Sam Slater, Joe Thwaites, PlayStation Studios Creative Arts, and Housemarquians recording at Temppelinaukion Kirkko, as they talk about the music of SAROS. 🔆
youtu.be/f7k76BkegRo
2
1,332
May 1
We need a Tyr & Jerome cartoon 🤪
Congratulations to @Housemarque for the release of Saros! We’re excited to meet Arjun Devraj and explore the incredible new world of Carcosa ☀️
1
14
1,049
Apr 26
Excited for you all to meet Jerome Jackson this Thursday (30 April). Thanks to @PlayStation and @Housemarque for another epic opportunity. Enjoy! 🔥🎉💣 #saros
3
29
265
5,610
Apr 26
These people came to see a show! 🔥💣
3
5
26
1,748
Apr 20
An Earthset.
Apr 19
Only one chance in this lifetime…
Like watching sunset at the beach from the most foreign seat in the cosmos, I couldn’t resist a cell phone video of Earthset. You can hear the shutter on the Nikon as @Astro_Christina is hammering away on 3-shot brackets and capturing those exceptional Earthset photos through the 400mm lens. @AstroVicGlover was in window 3 watching with @Astro_Jeremy next to him.
I could barely see the Moon through the docking hatch window but the iPhone was the perfect size to catch the view…this is uncropped, uncut with 8x zoom which is quite comparable to the view of the human eye. Enjoy.
2
1
17
1,104
Apr 12
Next weekend @SupanovaExpo in Melbourne! Come through and let’s talk! #apexlegends #godofwar #hades #saros #wolfman #playstation
Apr 11
Next weekend, meet @benjipea at @SupanovaExpo #Melbourne.
Further info here: dmjconsultancy.co.uk/upcomin…
#BenPrendergast
#ApexLegends #GodOfWar #WolfMan #StarWarsResistance #Hades #LegoStarWars #CallOfDuty #DuneAwakening
#Australia
1
11
1,139
Apr 10
Strap in folks, Jerome Jackson is in the house(marque)! 🔥🎉 #saros
Set out as Arjun Devraj, a Soltari enforcer, to traverse and brave the lands of Carcosa in Housemarque's Saros.
Launching April 30 for PlayStation 5.
1
10
1,172
Mar 21
Not me in the background ‘stirring the pot’ ⬆️
3
12
355
11,463
Mar 9
Some wonderful fun with @johnnyjohnleary and Ngaire Dawn Fair, who nailed the Saturday Quiz. podcasts.apple.com/us/podcas… 🧠🎉
2
647
Feb 20
5E ⬆️
2
19
1,696
Feb 10
More boom! 💥
5
4
70
4,581
Feb 8
What I’ve been saying all along. We were taught AI principles in 1997, the tech has caught up to store huge data sets and make inferences. That’s it. But AGI isn’t this. We are witnessing diminishing returns. Get used to billionaires telling you they need “one more version”.
Feb 7
AGI is not coming.
We are nowhere near AGI. What we have today is inference, not learning.
Models get trained once on huge fixed datasets, then frozen. You ask questions, they remix patterns they already saw. Nothing updates. Nothing sticks. Talking to the model does not make it smarter. It does not learn from you. Ever.
Learning is still slow, expensive - and offline.
Look at self driving. You drive around a pothole, make a U turn, and come back. The car’s AI does not learn that you just solved that exact problem. It reacts the same way every time using sensors and rules. Do this 20 times a day and it still has zero memory that the pothole exists. It just re sees it. That is why edge cases never die. There is no local learning. No accumulation.
No 'oh yeah, I’ve seen this before'
LLMs work the same way. Tell it your name and it does not remember. The only reason it looks like memory is because scaffolding keeps shoving your name back into the prompt every time and sanitizing the output.
The model itself has no idea who you are and cannot learn from interaction. It is structurally incapable.
And the scaffolding is the worst part. It is pure duct tape. Just prompts on prompts on prompts around a frozen model. When something breaks, nobody fixes learning. They add another layer. Another rule. Another retry. Another evaluator model judging the first model.
So you end up with systems that are insanely complex but mentally shallow. Debugging is hell because behavior comes from hack interactions, not a learnable core. Tiny prompt tweaks cause wild behavior shifts. Latency goes up. Costs go up. Reliability goes down. None of this compounds into intelligence. It just hides the cracks.
Until we have real persistent learning and real memory inside the system, there is no AGI.
LLMs are not built for this. You cannot prompt your way out of it. You need a totally different architecture. Yann LeCun is right.
And even then, what architecture can actually learn online, store memory, and stay stable on today’s hardware?
Best case, maybe 5-10 yrs.
Right now it is all inference. It looks magical, but the emperor has no clothes. A lot of people see it. Almost nobody says it out loud.
12
1,339
Ben Prendergast retweeted
Jan 20
People debate whether AI will ever be able to create good art. The answer is no. It’s impossible. If AI progresses exponentially for 100 years it still won’t be able to create any art, even bad art. That’s because art by definition is an expression of the human heart. It’s a language that can only be spoken or understood by beings with rational souls. AI can create spectacle. It can create distraction. In that sense it can create “entertainment” in its cheapest and most worthless form. But you can only confuse the machine’s soulless bullshit with art if you have made yourself nearly as numb and soulless as the machine itself.
1,689
405
5,827
637,431
Jan 7
Amazon out here suggesting I don’t rehearse so loud.
BELTBOX 3.0: Vocal Dampener for Singers, Actors, Performers, Stress Relief. A Portable Warm-Up Room. amzn.asia/d/e3D723N
3
858
Jan 6
TLDR: “Someone stole my stolen bike!”
Jan 6
“Make your own prompts” isn’t advice. It’s basic integrity.
I’m honestly fed up.
Changing a few words, renaming the prompt, or slightly rephrasing it doesn’t make it yours, the idea is still the same, the vibe is the same, and the results are obviously similar.
And no, this isn’t about one or two people, and it didn't happen once!!!!
I’m not mentioning names because I don’t want to hurt anyone, but please… stop.
Creating your own prompts is actually easier than copying someone else’s work! Try it
1
29
1,820
Ben Prendergast retweeted
30 Dec 2025
Folks, heres a list of things not happening by 2030 that you can take to the bank:
-AGI or any AI at all
-Autonomous vehicles deployed generally
-Autonomous robots of any kind
-datacenters in space
All of these technologies currently rely on regression to work. The first 3 use regression to compute predictions. The last one relies on the regression compute to be practical. All automation up until this point more or less relies on regression. It is incremental, not revolutionary, and IT DOESN’T SCALE. We are seeing this now, as models have now consumed all of the data on the planet and still can’t quite do anything correctly. In small domains with lots of data, it can be practical. Nearly all automation we have today leans on small domains to remain tractable. Once the domain expands, the problem blows up. Variance is the death of regression, data be damned. And the future is more variance than trend.
Humans excel because they are not statistical. They are superstitious. Adaption is quick, but leaves many data points unexplained. Most people who are extremely productive go their whole lives strongly believing a whole host of things that are not true. It is precisely our lack of statistical rigor that makes us effective.
So when you try to recreate human superstition with statistics, compute will blow up and edge cases will snuff it out in the night.
It is a certainty that regressive methods are a dead end. A certainty. It is not practical to require 10,000 humans worth of energy to recreate 1 human output.
Now could someone discover how the mind works and the methods it uses? Absolutely. But no one is looking for it because they are getting paid millions a year to bark up another tree. And once discovered, then a roadmap must be built to get us to systems that can perform it. We must design new chips, develop new material systems, new compute structures, how to interface them with conventional computing, and on and on and on. It’s a decade job once the roadmap has been developed. We don’t even know where to start!
The sooner we abandon the bankrupt religious belief that AI will save us, the better we will be. Because every day we ignore reality is another day for our problems to metastasize.
AI will not solve our problems. Robots will not solve our problems. We must solve our problems ourselves. And we could start building a roadmap for that today.
224
379
3,071
162,627
23 Dec 2025
RIP Vince, amazing creator, video game godfather. Apex Father. Thoughts go out to his family. ❤️ bbc.com/news/articles/cx25rl…
2
5
89
1,803
Ben Prendergast retweeted
22 Dec 2025
Happy Holidays from Housemarque! ⚙️
26
120
1,365
38,360