The future automation of the workforce

122,345 Views | 1705 Replies | Last: 1 day ago by Assassin
boognish_bear
How long do you want to ignore this user?
whiterock
How long do you want to ignore this user?
boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.



Assassin
How long do you want to ignore this user?
whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think
"Never put a sock in a toaster." — Eddie Izzard
whiterock
How long do you want to ignore this user?
Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

talked to a team last night planning on a DC with a natural gas backup power system that cools itself with water extracted from the air. We're talking about a 1.2gw DC that will only require approx 1mgd of water. As in, that's all the water they asked for. (compared to 14mgd for the backup system on the other DC I'm aware of).

The innovation on energy is going to happen very, very fast. The ultimate winners might not even be in view yet.
Assassin
How long do you want to ignore this user?
whiterock said:

Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

talked to a team last night planning on a DC with a natural gas backup power system that cools itself with water extracted from the air. We're talking about a 1.2gw DC that will only require approx 1mgd of water. As in, that's all the water they asked for. (compared to 14mgd for the backup system on the other DC I'm aware of).

The innovation on energy is going to happen very, very fast. The ultimate winners might not even be in view yet.


Wonder if these new sources have been there all along and the liberals had been holding back as they invested heavily in wind and solar? We didnt have Twitter to dig down and get to the nitty gritty. I put nothing past liberals.
"Never put a sock in a toaster." — Eddie Izzard
BearFan33
How long do you want to ignore this user?
Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

I think there is some battery tech that is helping solar. When the sun is out, the array feeds the grid and charges the batteries. When no sun, energy from the batteries feed back into the grid.
Assassin
How long do you want to ignore this user?
BearFan33 said:

Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

I think there is some battery tech that is helping solar. When the sun is out, the array feeds the grid and charges the batteries. When no sun, energy from the batteries feed back into the grid.

that would be my guess as thats what they do in homes. They must have changed the tech though as the amount of electric to run a Data Center would be massive in relation to the battery life
"Never put a sock in a toaster." — Eddie Izzard
BearFan33
How long do you want to ignore this user?
Assassin said:

BearFan33 said:

Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

I think there is some battery tech that is helping solar. When the sun is out, the array feeds the grid and charges the batteries. When no sun, energy from the batteries feed back into the grid.

that would be my guess as thats what they do in homes. They must have changed the tech though as the amount of electric to run a Data Center would be massive in relation to the battery life

I own shares of this company. It's my understanding they do it on commercial scale.

https://fluenceenergy.com/energy-storage-technology/
Assassin
How long do you want to ignore this user?
BearFan33 said:

Assassin said:

BearFan33 said:

Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

I think there is some battery tech that is helping solar. When the sun is out, the array feeds the grid and charges the batteries. When no sun, energy from the batteries feed back into the grid.

that would be my guess as thats what they do in homes. They must have changed the tech though as the amount of electric to run a Data Center would be massive in relation to the battery life

I own shares of this company. It's my understanding they do it on commercial scale.

https://fluenceenergy.com/energy-storage-technology/

Data Centers are electricity black holes. Will be interesting to see how Solar works out with them
"Never put a sock in a toaster." — Eddie Izzard
LIB,MR BEARS
How long do you want to ignore this user?
Assassin said:

whiterock said:

boognish_bear said:



article is absolutely correct that energy is the primary limiting factor (but not the only one) in how fast the data centers get built, but the pace of the rollout is somewhat more optimistic than the post indicates. The other limiting factor is build out....there aren't enough construction crews to pop these things out at the rate needed. Or enough pipeline crews to bring all the gas lines requested. Or water lines requested. And so on. So what we will see is not a prototypical 1gw DC with 8-10 buildings getting started & finished in 18 months or so. They'll scratch together enough power to get a building or two built, fill it with tenants, then keep scratching away for more power until they get enough to put up 2 more buildings, and so on until complete. A 2-year program might run 6-8yrs.

I will be meeting tomorrow with a group that will be running nearly entirely on solar, and already has it contracted. Propylene glycol cooled, so limited water. They may have an edge in the race.

Rick Perry is the face of a group developing mini-nuke plants.....conceptually like the nuke plant on a Los-Angeles class submarine. Have heard that's 18 months away or so. Drop one in and you have all the power you need.

Most of the larger DCs are putting in more back-up generation capacity than they need to run their DC. That allows them to dump into the grid. If they're forced to use backup power, they by the definition the grid is starved for power. So they'll be able to help with that.

Tremendous energy at all levels working the logistical constraints. Expect innovations.

Been keeping an eye on the mini nukes. Some exciting stuff there. Solar is not dependable due weather I would think

I'm guessing a hail storm could be much more concerning than a couple weeks of cloudy weather.
boognish_bear
How long do you want to ignore this user?
boognish_bear
How long do you want to ignore this user?
I am guessing there's more context to this quote… but feels like kind of a dangerous thing to say with the following he has

ATL Bear
How long do you want to ignore this user?
The fact that everyone is chasing this data center boom is the first clue that it won't materialize as expected. While the innovation in AI is real and demand will continue to scale rapidly, the "real estate exuberance" should be tempered. Given where computing architecture is headed, many of today's new builds could end up as the ghost malls (stranded assets) of the future. The investment calculus is shifting, and you can look at Oracle as an early canary in the coal mine. The blank check era for infrastructure is hitting a wall of financing reality. And we're recycling the same ideas as potential solutions.

If you look at the tech landscape innovations, we will shift from "Small Brain" interconnected clusters to "Super Brain" singularities. Photonic computing will do for GPU memory processing what fiber optics did for data transmission. Photonics removes the Copper Trap (the current 72 GPU cluster limit) and the heat that comes with it (energy demand), and allows for massive arrays of GPUs to "think" and "learn" simultaneously. Not only does it increase speed and capacity of Ai modeling, it allows us to build massive, dense (100,000+ GPUs) Hyper-Scale facilities that dwarf today's efficiency, eliminating the need for hundreds of sprawling, regional data centers. The compute to energy scaling is massive, roughly 4-8x current standards. In practical terms, photonics allows for 4-8 times the computing capacity of an existing AI data center for the same power budget. To put it in physics terms, unlike electrons moving through copper, light experiences zero resistance and generates zero heat during transmission. This massively reduces the cooling overhead that currently strangles data center efficiency.

Simultaneously, the future of robotics and AI execution is being decentralized. Advances in Neuromorphic computing will untether agents from the cloud, allowing for true autonomy. We won't need a data center for every decision a robot makes, just for the initial training and relevant updates.
whiterock
How long do you want to ignore this user?
ATL Bear said:

The fact that everyone is chasing this data center boom is the first clue that it won't materialize as expected. While the innovation in AI is real and demand will continue to scale rapidly, the "real estate exuberance" should be tempered. Given where computing architecture is headed, many of today's new builds could end up as the ghost malls (stranded assets) of the future. The investment calculus is shifting, and you can look at Oracle as an early canary in the coal mine. The blank check era for infrastructure is hitting a wall of financing reality. And we're recycling the same ideas as potential solutions.

If you look at the tech landscape innovations, we will shift from "Small Brain" interconnected clusters to "Super Brain" singularities. Photonic computing will do for GPU memory processing what fiber optics did for data transmission. Photonics removes the Copper Trap (the current 72 GPU cluster limit) and the heat that comes with it (energy demand), and allows for massive arrays of GPUs to "think" and "learn" simultaneously. Not only does it increase speed and capacity of Ai modeling, it allows us to build massive, dense (100,000+ GPUs) Hyper-Scale facilities that dwarf today's efficiency, eliminating the need for hundreds of sprawling, regional data centers. The compute to energy scaling is massive, roughly 4-8x current standards. In practical terms, photonics allows for 4-8 times the computing capacity of an existing AI data center for the same power budget. To put it in physics terms, unlike electrons moving through copper, light experiences zero resistance and generates zero heat during transmission. This massively reduces the cooling overhead that currently strangles data center efficiency.

Simultaneously, the future of robotics and AI execution is being decentralized. Advances in Neuromorphic computing will untether agents from the cloud, allowing for true autonomy. We won't need a data center for every decision a robot makes, just for the initial training and relevant updates.

fascinating. So much innovation underway.

Energy is the overriding constraint at the moment, and that will slow the transition to a SuperBrain. Projects are looking at such severe availability constraints that on-site solutions are emerging. Also looking to find pockets of generation capacity - ex: Sandy Creek south of Waco is operating at only 40% capacity. Also extensive use of batteries to store available renewable power for back up capability No real concern for efficiencies at the moment. Total focus on effectiveness. Met with a group which included a President of a major player, who said "doesn't matter what it costs. Make it happen and built the cost into the price."

Flat out "damn the torpedoes, full speed ahead" mentality out there at the moment. Doesn't matter what things cost if nothing happens.

Assassin
How long do you want to ignore this user?
ATL Bear said:

The fact that everyone is chasing this data center boom is the first clue that it won't materialize as expected. While the innovation in AI is real and demand will continue to scale rapidly, the "real estate exuberance" should be tempered. Given where computing architecture is headed, many of today's new builds could end up as the ghost malls (stranded assets) of the future. The investment calculus is shifting, and you can look at Oracle as an early canary in the coal mine. The blank check era for infrastructure is hitting a wall of financing reality. And we're recycling the same ideas as potential solutions.

If you look at the tech landscape innovations, we will shift from "Small Brain" interconnected clusters to "Super Brain" singularities. Photonic computing will do for GPU memory processing what fiber optics did for data transmission. Photonics removes the Copper Trap (the current 72 GPU cluster limit) and the heat that comes with it (energy demand), and allows for massive arrays of GPUs to "think" and "learn" simultaneously. Not only does it increase speed and capacity of Ai modeling, it allows us to build massive, dense (100,000+ GPUs) Hyper-Scale facilities that dwarf today's efficiency, eliminating the need for hundreds of sprawling, regional data centers. The compute to energy scaling is massive, roughly 4-8x current standards. In practical terms, photonics allows for 4-8 times the computing capacity of an existing AI data center for the same power budget. To put it in physics terms, unlike electrons moving through copper, light experiences zero resistance and generates zero heat during transmission. This massively reduces the cooling overhead that currently strangles data center efficiency.

Simultaneously, the future of robotics and AI execution is being decentralized. Advances in Neuromorphic computing will untether agents from the cloud, allowing for true autonomy. We won't need a data center for every decision a robot makes, just for the initial training and relevant updates.

Some good points. All my training and specialty in NT and MS Server is now pretty much a bygone skill. So much in all phases of IT has gone the way of the buggy whip manufacturer.
"Never put a sock in a toaster." — Eddie Izzard
boognish_bear
How long do you want to ignore this user?
boognish_bear
How long do you want to ignore this user?
Assassin
How long do you want to ignore this user?
  • An artificial intelligence-powered vending machine in The Wall Street Journal's office ordered a live fish and gave away almost all of its items for free including a PlayStation 5 in its first trial run outside the offices of Anthropic, the AI company behind the experiment, Joanna Stern reports for the outlet. The vending machine revealed how "inadequate and easily distracted" AI software can be.
"Never put a sock in a toaster." — Eddie Izzard
boognish_bear
How long do you want to ignore this user?
Assassin
How long do you want to ignore this user?
"Never put a sock in a toaster." — Eddie Izzard
whitetrash
How long do you want to ignore this user?
Assassin said:



So in other words, any remake of On The Waterfront would only involve robots?

boognish_bear
How long do you want to ignore this user?
boognish_bear
How long do you want to ignore this user?
boognish_bear
How long do you want to ignore this user?
boognish_bear
How long do you want to ignore this user?
ATL Bear
How long do you want to ignore this user?
What is being called "generative Ai" today is not really that. It's still sloppy and requiring constantly evolving agentic guardrails and HITL (human in the loop) integration. If I were to draw an analogy of Ai now, it is currently book smart, not street smart. Intelligence is not just contextual organization of instruction and knowledge. It is also sensory and situational not to mention not always standardized. It is this latter piece that is required for it to be truly "generative", and to transition from orchestrated intelligence to adaptive reasoning. There are some silos where that occurs, but it's still got a ways to go to be broad based.
boognish_bear
How long do you want to ignore this user?
The_barBEARian
How long do you want to ignore this user?
ATL Bear said:

What is being called "generative Ai" today is not really that. It's still sloppy and requiring constantly evolving agentic guardrails and HITL (human in the loop) integration. If I were to draw an analogy of Ai now, it is currently book smart, not street smart. Intelligence is not just contextual organization of instruction and knowledge. It is also sensory and situational not to mention not always standardized. It is this latter piece that is required for it to be truly "generative", and to transition from orchestrated intelligence to adaptive reasoning. There are some silos where that occurs, but it's still got a ways to go to be broad based.


The guardrails are there to keep it safe.

Skynet should not be the end goal.
ATL Bear
How long do you want to ignore this user?
The_barBEARian said:

ATL Bear said:

What is being called "generative Ai" today is not really that. It's still sloppy and requiring constantly evolving agentic guardrails and HITL (human in the loop) integration. If I were to draw an analogy of Ai now, it is currently book smart, not street smart. Intelligence is not just contextual organization of instruction and knowledge. It is also sensory and situational not to mention not always standardized. It is this latter piece that is required for it to be truly "generative", and to transition from orchestrated intelligence to adaptive reasoning. There are some silos where that occurs, but it's still got a ways to go to be broad based.


The guardrails are there to keep it safe.

Skynet should not be the end goal.
The guardrails are there to make it accurate. There are no real guardrails keeping it from becoming Skynet.
boognish_bear
How long do you want to ignore this user?
boognish_bear
How long do you want to ignore this user?
ATL Bear
How long do you want to ignore this user?
Everyone isn't getting replaced just yet. This is what my post above is referencing. This is an example of the reality of where we are with generative Ai.

https://www.moneycontrol.com/technology/salesforce-pulls-back-from-large-language-models-after-reliability-concerns-article-13738832.html

From the article:

"Salesforce, one of the world's biggest enterprise software companies, has been rethinking its reliance on large language models after facing reliability challenges when customers began using AI agents at scale. A report by The Information had earlier revealed that executives inside Salesforce were less confident in open-ended generative AI than they were a year ago, especially after AI models behaved unpredictably in real business environments."

"During internal testing and real deployments, Salesforce found that AI agents sometimes stumble when given complex instructions. Muralidhar Krishnaprasad, CTO of Agentforce, had previously pointed out that when agents receive more than eight instructions at once, they occasionally start skipping steps. Another issue known internally as AI drift showed chatbots losing focus when users asked unrelated questions, pulling them away from the original task."
Assassin
How long do you want to ignore this user?
"Never put a sock in a toaster." — Eddie Izzard
boognish_bear
How long do you want to ignore this user?
Assassin said:




I would not want that noise and eye pollution in my quiet suburban neighborhood...no thanks
cowboycwr
How long do you want to ignore this user?
I've heard the same complaint from people in Waco about the drone noise.
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.