Thursday, November 6th 2025

Microsoft's AI Goal is Humanist Intelligence in Service of People and Humanity

Microsoft's head of AI Mustafa Suleyman made some bold promises in which direction the company's AI research will be headed. "At Microsoft AI, we're working towards Humanist Superintelligence (HSI): incredibly advanced AI capabilities that always work for, in service of, people and humanity more generally," notes the lengthy blog post, adding that "We think of it as systems that are problem-oriented and tend towards the domain specific. Not an unbounded and unlimited entity with high degrees of autonomy - but AI that is carefully calibrated, contextualized, within limits. We want to both explore and prioritize how the most advanced forms of AI can keep humanity in control while at the same time accelerating our path towards tackling our most pressing global challenges."

To achieve these goals, Microsoft has built a team called Microsoft AI (MAI) Superintelligence Team. Using massive amount of resources, combining human intelligence, hardware, software, and various other intelligence forms, Microsoft wants steerable AI systems. However, there is an issue. It appears that no one knows how will a superintelligent system be controlled. As Mustafa Suleyman claims, "No AI developer, no safety researcher, no policy expert, no person I've encountered has a reassuring answer to this question. How do we guarantee it's safe?"
Instead of having a goal to surpass human at everything and take over, MAI Superintelligence team wants Humanist Superintelligence (SAI) to be the one that aids humans instead of replacing them. "Humanist superintelligence (HSI) offers an alternative vision anchored on both a non-negotiable human-centrism and a commitment to accelerating technological innovation… but in that order. The order is key. It means proactively avoiding harm and then accelerating. Instead of being designed to beat all humans at all tasks and dominate everything, HSI begins rooted in specific societal challenges that improve human well-being." Let us know your opinions about this, and whatever you believe Microsoft could build a human-centric AGI.
Sources: MAI, via The Verge
Add your own comment

53 Comments on Microsoft's AI Goal is Humanist Intelligence in Service of People and Humanity

#26
lambda
It seems they will work on hyper specialized models with a safety first approach.

I hope they opensource it, gpt-oss-20B style models tuned for specific domains would be quite nice.

The rest of the announcement is pure PR.
Posted on Reply
#27
kondamin
Jacky_BELDoes this mean they will have to outsource learning their AI the concept of Ethics , since MS seems to have lost the way.
Lost their way?
Microsoft hasn’t historically been the nicest of companies.

It’s been better behaved than say united fruit, union carbide and Monsanto but I wouldn’t say it’s ever been a nice and friendly or a producer of rock solid software.
Posted on Reply
#28
DeathtoGnomes
I'm just not gonna say it, its loud enough as it is.
Posted on Reply
#30
Tek-Check
Ah... now I see. It is Microsoft that wants Nobel Peace Prize.
Posted on Reply
#31
Vayra86
freeagentI dunno man..

This whole AI thing seems to be a bad idea.

Call me old fashioned :rolleyes:

Slowly, the paradoxical reality of the AI is coming to life.

- Either you control it meticulously and it quickly becomes a very expensive Clippy
- Or you let it go free and it will push the emotional to suicide, the average human to the mental asylum, and itself to pointlessness, while it eventually still reasons itself in a deep hole, nonetheless.

Sentience is far away. Its still a computer, and its still completely an affair of 'input = output'. AI doesn't do fuck all by itself. Its all human input, in the end.

Studies are showing that every AI you repeatedly ask similar questions, will shortcut its way to earlier answers. Minutely different, but similar, and not necessarily 'correct'. Basically the AI suffers from confirmation bias, just like humans do. Yet another strike against the AI: it reasons like humans do, because human efficiency dictates we aren't going to fix the same problem again, we just look back on how we did it before, AI is just as lazy as a human, at that, too. And it lacks the human drive, a motivator to keep exploring - so its effectively worse.

The bottom line, AI ain't better, its just a bigger dataset and it can distill something out of it. But in its practical use, its no more than an overpriced algorithm and once you've got the right one, you don't need the AI to do any learning anymore (but you could still call it 'AI'). This state is what Microsoft is describing here and I think its the only logical 'out' to somehow recover a snippet of the billions spent on it: distinct Agents that execute specific tasks - very expensive, but specialized Clippy's.
Posted on Reply
#32
AVATARAT
Vayra86The bottom line, AI ain't better, its just a bigger dataset and it can distill something out of it. But in its practical use, its no more than an overpriced algorithm and once you've got the right one, you don't need the AI to do any learning anymore (but you could still call it 'AI'). This state is what Microsoft is describing here and I think its the only logical 'out' to somehow recover a snippet of the billions spent on it: distinct Agents that execute specific tasks - very expensive, but specialized Clippy's.
Like any other tool, AI will need time to be refined, which is one of the reasons why it is being pushed so aggressively everywhere—to improve it faster.
The main purpose of such a tool is that it can do things much faster and 24/7, which saves or will save money for large companies.
Just look at automated car factories as an example, some things are still done by humans, but many operations are automated, it will be the same here.
Another example, did you know that there are Microsoft 365 companions? You and I won't use it, but most people will just accept it. And they will improve this type of AI Apps, Linux will be an option, but for how long?
Posted on Reply
#33
theouto
Great words, and you formed some great sentences with them. Good job Microsoft! You can write a good blog-post, shame that those words are absolutely meaningless and you know damn well that you are mostly talking drivel, none of what I just read makes any sense outside of an imaginary land where what they are building is more than a predictive algorithm.

If only said drivel didn't make line go up...
Posted on Reply
#34
TumbleGeorge
AVATARATLike any other tool, AI will need time to be refined, which is one of the reasons why it is being pushed so aggressively everywhere—to improve it faster.
The main purpose of such a tool is that it can do things much faster and 24/7, which saves or will save money for large companies.
Just look at automated car factories as an example, some things are still done by humans, but many operations are automated, it will be the same here.
Another example, did you know that there are Microsoft 365 companions? You and I won't use it, but most people will just accept it. And they will improve this type of AI Apps, Linux will be an option, but for how long?
Totally from moment that "AI" mania starts to end of 2025 spendings for "AI" probably will touch $1.5T...With perspective for much more for next few years. When positive incomes from "AI" will become to reality? Never!
Posted on Reply
#35
Vayra86
AVATARATLike any other tool, AI will need time to be refined, which is one of the reasons why it is being pushed so aggressively everywhere—to improve it faster.
The main purpose of such a tool is that it can do things much faster and 24/7, which saves or will save money for large companies.
Just look at automated car factories as an example, some things are still done by humans, but many operations are automated, it will be the same here.
Another example, did you know that there are Microsoft 365 companions? You and I won't use it, but most people will just accept it. And they will improve this type of AI Apps, Linux will be an option, but for how long?
Cryptocurrencies, like any other tool, needs time to be refined. I mean yes, it uses an ungodly amount of power and resources to automate something we could already do. But its so disruptive, soon, you will want nothing but payments in bitcoin! After all, we are democratizing finance here, power to the people, YOU are in control! Yes, I know you cant see it yet, but believe me. Soon!


Autonomous driving, much like any other tool just needs time to be refined! Yes it does something we already did and takes away a skill we used to need. But just wait! Soon! We already have cab rides in a controlled environment and sure its still a little problematic but just wait! Soon!
:toast:
Posted on Reply
#36
Aquinus
Resident Wat-man
freeagentI dunno man..

This whole AI thing seems to be a bad idea.

Call me old fashioned :rolleyes:

I don't know if it's bad, but it definitely feels like a bubble that's going to pop, then it will be bad.
Posted on Reply
#37
Darmok N Jalad
AquinusI don't know if it's bad, but it definitely feels like a bubble that's going to pop, then it will be bad.
I can’t recall seeing something so heavily invested in before. Maybe there has been, but this one feels different. With the crypto boom, these trillion dollar companies weren’t busy building massive data centers for a specific task. Maybe these facilities can be repurposed, but if AI pops, many billion dollars will be lost by each of them.
Posted on Reply
#38
TumbleGeorge
Darmok N Jaladmany billion dollars will be lost by each of them.
That's not sound bad, I don't know why. .
Posted on Reply
#39
AVATARAT
TumbleGeorgeTotally from moment that "AI" mania starts to end of 2025 spendings for "AI" probably will touch $1.5T...With perspective for much more for next few years. When positive incomes from "AI" will become to reality? Never!
I can only speculate based on what I read, so probably between 3 to 10 years.
And I wouldn't dare say that every technology company has so many stupid people in it that they would bet billions of $ on a "bubble."
And why do you think that "positive income" can only be "money"? ;)
Vayra86Cryptocurrencies, like any other tool, needs time to be refined.
CBDC + AI management? :)
Posted on Reply
#40
TumbleGeorge
AVATARATAnd I wouldn't dare say that every technology company has so many stupid people in it that they would bet billions of $ on a "bubble."
Their money or investors money? I have no problem to spend trillions investors money in baloons.
Posted on Reply
#41
EsaT
...most pressing global challenges.
Most pressing global challenges are all the ongoing destruction of environment and various pollution.
Now how does wasting lots of electricity and all the hardware environmentally expensive to make from raw materials help to solve those?

Well, I guess if it's to make Skynet which then curbs this nature's top pest specie...
Posted on Reply
#42
Darmok N Jalad
EsaT...most pressing global challenges.
Most pressing global challenges are all the ongoing destruction of environment and various pollution.
Now how does wasting lots of electricity and all the hardware environmentally expensive to make from raw materials help to solve those?

Well, I guess if it's to make Skynet which then curbs this nature's top pest specie...
they’ll get this done and come back with “42” as the meaning of existence.
Posted on Reply
#43
rutra80
SOMA comes to mind when thinking about where does the road paved with good AI intentions lead to...
Posted on Reply
#44
boidsonly
It's no coincidence M$ bought 27% of Open AI/ChatGPT.
Posted on Reply
#45
ypsylon
Said one of the most obnoxious and corrupt corporations in recent history...

Certainly a lot of (poop) weight in M$ statement.

The sooner this AI cancer blows up the better.
Posted on Reply
#46
notoperable
It seems at Microsoft, AI Hallucinations spread to the staff
Posted on Reply
#47
remixedcat
Microsoft can't vibe code their way out of a paper sack
Posted on Reply
#48
notoperable
ypsylonSaid one of the most obnoxious and corrupt corporations in recent history...

Certainly a lot of (poop) weight in M$ statement.

The sooner this AI cancer blows up the better.
Please, look on Billy boy's sudden PR remake in a decade- from corporate megalomaniac/psychopath to humble wool sweater x-mas grandpa good doer unless its more then 1K$ in flat costs :-) Lets give M$ a chance, maybe they drop their SaaS (System as a Sypware) data mining on planetwide and won't reduce humans to data-points but will be all good for all of the humanity (i wonder what their interpretation of humanity is) - Lets give them a benefit of a doubt. We gonna colonize Mars frist before M$ is gonna do anything for anyone without a balance sheet.
remixedcatMicrosoft can't vibe code their way out of a paper sack
Mark my words, next thing good ol' monkey-boy Ballmer is gonna be recalled from Basketball trolling to run M$ properly again. I see a lot of potential for shareholders in Steven's reinstation in those difficult times for their profitbag

I thing Goebbels is more credible then any effort from M$ PR spin-doctor carousel that spits those grandiose mission statements since 1995 without any shame
Posted on Reply
#49
remixedcat
notoperablePlease, look on Billy boy's sudden PR remake in a decade- from corporate megalomaniac/psychopath to humble wool sweater x-mas grandpa good doer unless its more then 1K$ in flat costs :-) Lets give M$ a chance, maybe they drop their SaaS (System as a Sypware) data mining on planetwide and won't reduce humans to data-points but will be all good for all of the humanity (i wonder what their interpretation of humanity is) - Lets give them a benefit of a doubt. We gonna colonize Mars frist before M$ is gonna do anything for anyone without a balance sheet.


Mark my words, next thing good ol' monkey-boy Ballmer is gonna be recalled from Basketball trolling to run M$ properly again. I see a lot of potential for shareholders in Steven's reinstation in those difficult times for their profitbag

I thing Goebbels is more credible then any effort from M$ PR spin-doctor carousel that spits those grandiose mission statements since 1995 without any shame
Even tho Steve ballmer was a sweaty asshole he did a much better job of running ms than salty nutella is doing rn!
Posted on Reply
#50
Vayra86
remixedcatEven tho Steve ballmer was a sweaty asshole he did a much better job of running ms than salty nutella is doing rn!
Developers developers developers!!

Oh man what a dude. You know, Steve's persistence might have leaked into Windows' update schemes and will never go away, like a piece of rogue code it just keeps trying to wedge its way into your life. Its the only possible explanation for all those applications that keep trying to set themselves as default...
Posted on Reply
Add your own comment
Nov 21st, 2026 13:02 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts