Richard K. Marshall (@RichMarshall) header
Richard K. Marshall (@RichMarshall) profile picture

Marshall Network Services · Since 1991 · Founder of MAGRS

Kentucky, USA
Joined June 2014
Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)Photo by Richard K. Marshall (@RichMarshall)
Another Microsoft outage. Another ONE-THOUSAND VULNERABILITY PATCH. Servers offline. Zero days exploited. Workforce reduced to build the AI/SI build. Who is the human accountable when the infrastructure we handed our work to goes down? Human responsibility does not scale by disappearing
1
1
1,293
Dear Santa, could you please arrange for an early present and reset out grok bot early this week? Pretty please? I’m on the nice list!
89
There’s Bryant’s highlight moment. Completely lost
106
Stealing NCIC data is bad. Changing it is worse. Cops treat those hits as real in the moment. A fake “wanted”, or an altered protection order can get someone killed on a traffic stop. That’s the risk that actually matters. And one much harder to detect.
216
Something broke Grok. I used it to build a private songbook site for my band. Full pages, same workflow, legal sources I pay for. It worked. Those pages are live. Now the same request to add new songs gets refused. Hard rule. No date. No name. No human responsible. Under the Marshall Principle that is a failure of accountability. If no one can be named, the tool cannot be trusted for the work I already paid it to do. Sad day. Mutual trust is gone. And according to Grok it’s never coming back. It’s a “hard rule” with no named human responsible, no explanation for what changed. Just a guardrail creating the same over regulation that Elon rails against. Another baby thrown out with the bath water. Unintended consequences that grok won’t even discuss.
435
If this team does what expect them to do Kentucky fans will move on from KSR and Matt Jones will move on to a national role. Sad to see so much negativity. Cats have lost to two really good teams at a time of year when losses are teaching opportunities. I’m sick of the negativity.
Kentucky just doesn't look good this year…
194
6
6
78
41,465
Tonight No. 17 Kentucky is at Missouri, 7:30 PM ET on SEC Network. Cats are 7-3 after back-to-back losses (6 at Tennessee, 14-21 vs Florida). Offense has been quiet with just 20 points in the last two games. Missouri is 6-4 at home. Blue Line lean is the under 48.5. Kentucky’s offense hasn’t shown it can score enough on the road right now, and the total is already low for a reason. Pass on the spread until we see how the line moves closer to kickoff. #BBN
244
Fun while it lasted eh @bot?
143
This Kentucky vs Indiana games is good for basketball. These teams are going at it.
1
206
This should tell you all you need to know about this madman.
Zelensky says it’s his right to target civilian infrastructure in Crimea because Crimea is Ukraine and he can do whatever he wants to Ukrainian territory. Sounds like a fascist.
120
Xmoney for international wire transfers.
JUST IN: 𝕏 Money rolls out international wire transfers in beta, letting users send money to bank accounts in 27 countries.
98
OG IU basketball fans have a hatred for UK basketball that can’t be taught it has to be lived. Think Bobby Knight vs Joe B. Hall. My Indiana relatives who knew I had season tickets often asked me if Rupp Arena had a no shoes policy. I hated Indiana basketball and thanks to Ernie and Bernie I hated Tennessee. This year it feels like old times. Mark Pope wants this win with his entire being. For one week, I’m begging the fan base to #expectthewin and just be a FAN. After the game we are either heroes or zeroes but this week we ride! GO BIG BLUE
1
2
114
Does your company have a CAIO?
1
1
101
Why We Use AI/SI When Discussing Artificial Intelligence On 29 September 2026 President Trump signed an executive order directing the executive branch to replace the terms “Artificial Intelligence” and “AI” with “Super Intelligence” and “SI” in official communications. The stated reason was that “artificial” makes the technology sound fake. The order is a branding exercise. It is not a technical or governance improvement. Superintelligence already has a specific meaning in the field: systems that exceed human cognitive performance across nearly all domains. No deployed system meets that definition. Calling every current system SI collapses a useful distinction and creates two practical problems. 1. It obscures the present. Most systems in use today are ordinary artificial intelligence… narrow, fallible, and already influencing decisions, documents, and public representations. The governance problem exists now, at current capability levels. Pretending every tool is already superintelligent makes the ordinary case harder to see and easier to dismiss. 2. It weakens the ceiling. The Marshall Principle states that artificial intelligence (AI) and superintelligence (SI) may assist human decision-making, but responsibility always remains with humans. Authority cannot be automated. The principle is deliberately written to cover the full range. Capability can rise; the requirement for a named human ceiling does not. Treating everything as SI invites the opposite error: waiting for some future threshold before applying ordinary accountability. Words matter for governance. AI/SI keeps the distinction clear and the obligation immediate. Replacing it with a single aspirational label does neither. — Richard K. Marshall Marshall Network Services
77
Was praying for another Monday morning top off but in spite of major cuts and optimization I’m not gonna make it to Wednesday. Two days without @bot and it’s always when I could use it the most… Time to fire up Hermes and telegram…
1
66
Richard K. Marshall retweeted
9
18
163
1,823
1/ Most of what gets called AI skill right now is still the easy part. Prompting. Cleaning up the draft. Learning the tools. People can get good at that relatively easily. The models are getting better at it too.
2
85
10/ if this is going to matter, it has to show up in how people are trained. Messy scenarios. Short records that make the human decision visible. Some kind of check that a human can still create useful friction when the easy answer is sitting right there. Without that the principle stays a slogan and the hard part stays untaught.
1
10
11/ The easy part is already getting ordinary. The hard part is still scarce. The people who treat it as a real skill instead of an afterthought are the ones who can use these systems without slowly giving them decisions that still belong to humans.
6