Musk, Meta, OpenClaw, MoltBook — and the Fight for the Future of Work

00:00:00: In a lot of ways what he's describing is how we will be interacting with AIs in the future.

00:00:06: It used to be, you know... We would type into create a symbolic system that describes where we want program it and then it runs.

00:00:15: And last few years have been chatting about these systems and you give it a question that comes back with the response.

00:00:23: The agent systems, uh...that are coming now essentially allows us to be able to hand off a task for them And then they go out into work.

00:00:32: Then if you have team of them who can talk to each other You have manager agents talking about the team of agents which has an ability To create company inside computer.

00:00:46: Welcome to Big Bang Tech Report!

00:00:48: Today the next AI war.

00:00:52: We will not question who has the smartest chatbot, let's talk about Who wins control?

00:00:58: Musk, Meta, Openclaw, Moldbook, NVIDIA, Newchips, Nemoclaw and Jensen Wang is calling Openclaws the operating system for the Agenteca H. Wow there are a lot of stuff!

00:01:14: So...who will control?

00:01:17: Alvin Greiland is with us.

00:01:19: Alvin, great to have you here!

00:01:22: Yeah it's good to see again as a lot has happened in the last two weeks since we talked.

00:01:25: yeah

00:01:26: It's so incredible.

00:01:28: So fast start Alvin.

00:01:30: are We now moving beyond chatbots and into the age of AI systems that will work for Us.

00:01:36: let's Start With Musk terrafeb.

00:01:39: what Is That?

00:01:43: I mean i think one Of The big news that Happened In the Last few Days was Musk's announcement that he is making a terawatt chip factory so, That he can supply enough chips because He doesn't think that TSMC and Samsung And all the other companies Intel Can create Enough Chips for him.

00:02:03: right I Think in some ways.

00:02:06: He Is probably Right In The Near Term There Is A Supply Constraint.

00:02:11: But i feel like Overestimating his own ability to deliver these things because it's essentially taking decades for these companies To be able to Create as many products as they have today and he is essentially trying to create more than the capacity of all Of the semiconductor suppliers in the world In the next two or three years which It's not only not feasible.

00:02:41: It's in fact that type of a goal and the type of things he is trying to do, it will probably make him detrimental for his ability to deliver because...

00:02:52: But tell us what are you really building?

00:02:54: A chip empire or an eye-empire Or both?

00:02:58: What does this

00:02:59: mean?

00:03:00: Well He was trying to create an entire vertical stack Of every single component materials, silicon to chips.

00:03:11: To the tools that make the chips and robots.

00:03:17: they will go into space systems which launch it then energy systems in sky that bring terawatts of energy.

00:03:31: So essentially he's trying create an entire global economy with one company and supposedly for twenty five billion dollars, which you know it's the number that he is talking about in a time frame.

00:03:46: With the scope of what his creating there just isn't an impossibility to happen.

00:03:53: I'm not saying this idea longer term we expand capabilities on these areas but for a company that has never done semiconductors to go into semiconductor.

00:04:05: I used to work in semi-conductor companies and know how hard it is, the time frames.

00:04:10: they worked because fab takes five or six years for him to say within three or four years he's going to exceed the global capacity of the world and then, in order to be able to send this up into space.

00:04:30: He would need thousands actually tens of thousands hundreds of thousands of starship flights every year To send this material up.

00:04:39: so far we've had something like five or six flights And half them have blown out the reality and what he's proposing, there is a very big gap.

00:04:53: But it can work.

00:04:54: if every Tesla can fly in few years you know?

00:04:57: Then It Can Work!

00:05:02: I think If You Look Underneath Because He Is A Smart Guy And I Think That He Has Also Done The Math And He Knows This Not Possible but the Reality of Why He Is Saying These Things Actually Much Simpler.

00:05:13: and it's that he has an IPO coming up for his SpaceX.

00:05:19: He is now integrated SpaceX with XAI, so the combination of two creates a very strong story.

00:05:28: The need to make Space Data Centers create a reason.

00:05:32: there are thousands or hundreds of thousands flights for it happen.

00:05:39: So it's a fiction that allows for an evaluation to be justified beyond what is probably should and could be, but its not a reflection of reality.

00:05:52: Okay!

00:05:52: That brings us more or less to NVIDIA.

00:05:55: at GTC Jensen pushed new ships.

00:05:58: naval claw in the big idea open claw can become operating system for the agetic age.

00:06:07: What do you think about this?

00:06:09: I mean, actually admire... Jensen in so many ways, and he's very much ahead of the game for several different generations of technology.

00:06:20: And I think that a lot of what he is describing it how we will be interacting with AIs in future It used to be you know We would type into A system create a symbolic system That describes where we want program it in and then runs.

00:06:37: The last few years has been.

00:06:38: we are chatting With these systems And you know, if you give it a question that comes back with response the agent systems That are coming now essentially allows us to be able to hand off A task them and then they go up into work.

00:06:52: Then If You Now have a team of Them They can talk To each other?

00:06:55: And then you Have a manager Agent that Then Talks With The Team Of Agents.

00:06:59: You Essentially Have the ability to Create A Company Inside A computer.

00:07:05: right and I think that's what What he's Describing and i Think That Actually Will Be in a lot of ways, the interface model that we have with AI.

00:07:14: So what he's describing I think it actually underplays because an operating system still means like Linux or iOS and Windows where you're just using to get response and use tools inside.

00:07:30: What i actually thinks is not an Operating System for computer.

00:07:34: It's actually an Operated System for digital economy And that digital economy can be very autonomous in terms of how it interacts with each other, and these master agents can actually become the CEOs of their teams of AI.

00:07:51: So you became this shareholder of these virtual digital companies?

00:07:58: I was just in Austin have attended South by South where people talked a lot about Alton Claw But they were a little bit afraid of it and said, OK we buy new computer with our own stuff not everything.

00:08:14: And then we tested on the computer.

00:08:16: why do you that?

00:08:21: Open Claw is amazing because it's so open.

00:08:24: It allows you to do anything that can be done as a human on the computer and reach out any model, but by doing this makes it very vulnerable for security issues has access to your full computer, and it has access you bank accounts in all of the email.

00:08:45: And then hard drive.

00:08:46: they could delete everything that can send message out to friends take a go-to-bank and transfer money.

00:08:52: so there's definitely lot risk doing that.

00:08:55: some people are essentially creating kind sandbox devices limited their capabilities.

00:09:03: what was really happening last few weeks is every company now saying I recognize this vulnerability a defanged version of OpenCloud, which is actually what NemoCloud.

00:09:14: What

00:09:15: does Nemo Cloud explain

00:09:16: to us?

00:09:17: In the sense you're saying hey I'm going create in mini-version of open cloud that has its own virtual sandbox and it's safe.

00:09:24: in the sense if you have to give access to different tools so now they cannot do harm... It's less dangerous but not dangerous you know, full access like the open-class system.

00:09:37: Somebody at South by South told me that he is afraid his computer will call in five times and we'll tell him something what's not all right.

00:09:47: so I think they're strange.

00:09:50: In fact some people are using it for this way Right?

00:09:52: They essentially have a tie to their WhatsApp or signal or telegram and go do tasks through these messaging systems when they finish your tasks or when they have an update.

00:10:06: So in some ways, you are actually interfacing with it like You Are With A Friend Or A Colleague Right?

00:10:12: It's not necessarily a bad thing for Your AI to proactively call you because what that does is That means you don't Have To Sit At The Screen Anymore!

00:10:20: You Don'T Have To Say Okay Is It Running?

00:10:22: And They're Improving Every Single Piece.

00:10:24: What Do You Recommend?

00:10:26: What Shall I do?

00:10:27: We Do With Open Closhed Shall we use Nemo Claw or what is your advice?

00:10:33: Or shall?

00:10:35: So so I actually ran open cloud for about a week or so and And it was a lot of work to set up in a lot Of work to maintain.

00:10:42: Yeah, what i'm finding right now For me personally?

00:10:45: I can do everything that I was doing on OpenClock using clock code and Using the systems that is provided and its much safer.

00:10:53: It's a tool That already have an hour they haven't give it access To certain folders on my computer and with The new tools and the deep dispatch capability using my local computer.

00:11:06: So it's already doing a lot of the things.

00:11:08: I don't know if everybody needs to go and spend, you know day or two this setup at Open Claw.

00:11:13: One more day?

00:11:16: Okay yeah I think if you to do it right and with the current systems out there, It is not designed for the average user.

00:11:23: This why we talked about a few weeks ago or couple of weeks ago.

00:11:27: in China people were lining up To have engineers set-up.

00:11:30: they had these elderly grannies And grandpas lining up too.

00:11:34: get open clock as they were told that this Is going.

00:11:37: make them you know, thousands of dollars a week.

00:11:41: And what's actually happening right now is that these... A lot of these people who got them set up are not going back and saying can you uninstall it for me because I don't know if taking too many tokens is costing me tens of dollars per day of tokens?

00:11:54: But maybe later on this much more easier.

00:11:57: Is It Right Now a real turning point to the new future?

00:12:01: This New Systems

00:12:02: Action fuses.

00:12:04: I think it is actually, the usage metaphor of in the past us having a real-time interaction with AI and using typing and words by themselves text as an interface that's about to change make much more conversational and asynchronous And much more task based instead information base.

00:12:28: So from this perspective I would recommend all of the audience to go and actually experiment with it, understand that because then you will know kind how the interaction models are going change over time.

00:12:43: And your relationship with computers or cognitive work is changing.

00:12:49: So essentially if we can start thinking about these agents as either You know, an intern or a junior staff and how you would interact with it.

00:12:59: That's how these systems are working.

00:13:01: And then there is somebody something else showing up moldbook We have to add to the picture.

00:13:08: and what do you think about more book?

00:13:10: I mean yeah, and in the news when mobile recently was that It was acquired by face.

00:13:16: Yeah right or buy.

00:13:17: but uh Honestly i mean mode book which was something that was vibe coded in a day When people went back and looked, you know even though they said there was a million agents on their what it looked like.

00:13:32: There was at the end when people did.

00:13:33: that will digging is that there's probably about fifteen hundred independent machines accounts from different people that were running on it, and a lot of the conversations that we're very hyped up where actually orchestrated by humans to create PR.

00:13:48: So I think whatever large high price Zuckerberg paid for there was probably overpriced but its an interesting concept because in time agent-to-agent communication is going be thing will happen.

00:14:03: But it doesn't necessarily need this type of social network model for communication.

00:14:09: It's really that's something I think the community.

00:14:13: That reads too much sci-fi right now is Climbing on top because it really gives a sense at these.

00:14:19: AIs have Right Now existing consciousness and they are their own ascension beings, and they have your own will.

00:14:27: I'm not saying that won't happen And i'm not seeing as not possible theoretically but it feels like a lot of that imposing or some people, imposing their philosophies on top of what's happening.

00:14:38: And some of it is also humans injecting information to give the perception that happens.

00:14:44: So with a new players.

00:14:45: so within you AI agents we see change in real work.

00:14:51: We see change at the word desk email research scheduling presentations coding.

00:14:58: Where will people feel?

00:14:59: At first?

00:15:02: So I mean, one of the things that we had talked about in our exchanges on what's happened is that Zuckerberg just announced he now has his own sense chief of staff as an AI agent and making every team member do the same.

00:15:19: And so they have this collective system for using AI agents to manage their organizations creating a much thinner organization.

00:15:28: there almost no middle management Right.

00:15:31: And I think that they are definitely ahead of the game there and, you know having worked in very large organizations There's a lot of bureaucracy and inefficiencies That come from multiple layers Of management.

00:15:43: also The fact that CEOs last to know anything Because all people underneath Are filtering out what They want To hear instead What They need.

00:15:54: the approach that he's taking makes a lot of sense, and organizations that does adopt that type of model will probably see significant transparency in efficiency gains.

00:16:07: But what it also does is essentially adds to level full surveillance over all your employees because anything can now be searchable or auditable.

00:16:19: And if you say, okay well how much time did Yen's work on this project?

00:16:24: Did he do it himself or just ask the AI to do it.

00:16:27: Okay looks like ninety nine percent was done with AI.

00:16:29: maybe I don't need Yens anymore right.

00:16:31: and so This is kind of things that could be happening and managers are very profit driven or focus will start using a model for Using new information they're gaining which i think may actually create decisions that are more extreme than they need to be and faster.

00:16:55: In South by South, people will ask if they would prefer a human being or an AI as supervisor?

00:17:06: And what do you think was the answer?

00:17:09: mostly?

00:17:11: I mean... that South by Southwest is much more of a kind of creative and arts, human centric.

00:17:19: I would think they will want to be humans?

00:17:21: No it was the opposite because an agent is fair he has no bad moods He's reliable And so people say It could be good To have machine at the boss Because of transparency.

00:17:40: You know

00:17:44: A lot of what you say actually does make sense.

00:17:46: In fact, I have a new essay coming out soon on AI and leadership And i actually also advocate for more usage Of AI in helping You Make More Rational Decisions But to Have It Completely Be Offloaded.

00:18:01: Where Humans Are Now Reporting To AI We May Not be Ready For That Yet Because The Problem Is not the AI ,the problem is the person On top.

00:18:12: And so if I had Zuckerberg running an AI that was managing me, i would be very worried because the intent of the person on top is going to be reflected through The AI systems a lot more.

00:18:24: If objective function is to say make my organization flourish, help make people stronger.

00:18:32: Help develop them into more capable employees and professionals that's great.

00:18:38: if the objective function Is find a maximum efficiency in our organization tell me anybody That's not working hundred percent or hundred ten percent And then notify me so I can have them be laid off.

00:18:48: Then it's going to create A very different organization and The pressure of that second type Of scenario It's Going To Create a fear-based management system.

00:18:58: Do you think we will use it in the few years?

00:19:00: that is pretty normal to have an AI as a supervisor, or manager?

00:19:07: I think that AI being implemented and in the management process would become natural because of efficiency gains like what we just talked about efficiency of employees more than you've ever been able to.

00:19:28: So I feel like it's gonna be good for efficiency near term, and maybe a real issue for societal harmony and also the stress levels of employee long-term.

00:19:43: And so this is why we talked about before that there are going to an accelerating job displacement problem And that these type of management models is going to bring about that acceleration because the transparency of innate inefficiencies.

00:20:02: Humans have biologically probably two or three hours a day of clear attention span where we can work on something.

00:20:11: But if the system is monitoring, you're saying okay there was five minutes when he didn't do this and ten minutes were... And added together out at eight hours in office only had six point-five hours productive work.

00:20:24: You should reduce his pay by seventy percent.

00:20:27: These are kind things that an AI might tell manager but it's probably not necessarily long term beneficial because a lot of the productivity that happens inside humans is things when you're in peaceful off time, your creativity and creative mind can happen.

00:20:46: But if you are always working to try make next widget move it faster keep typing on keyboard so looks like you work.

00:20:56: You're not going to be creative and doing your best.

00:20:58: Yeah,

00:20:59: there was a common sense in Austin In Texas that we need people who question the AI Who asked the I if it is truth?

00:21:07: And that this is good Your story or message for people older than forty years because they used to know are they then no more about life About our circumstances and so on and that they will use That?

00:21:22: they would be very important for the economy.

00:21:26: It's not so good for young people, but it is good for older than forty years.

00:21:30: Do you agree with that?

00:21:33: Yeah I think there are always value in life experience because even though young people will be better at using these tools and they're going to have skills the older generation doesn't have.

00:21:47: But they ask how we can make them faster.

00:21:51: The older generations say why should do this?

00:21:56: And in a lot of ways.

00:21:57: where now the how becomes very automatic because the AI will have read every single manual and know it better than anyone, how to do.

00:22:06: The real question is why?

00:22:08: Should we be doing that?

00:22:10: recently with like the Iranian war Being put into the targeting system and an analytical systems, you know that news just came out That on a first day a thousand different targets were given to the military On what should be where the missile should go.

00:22:29: You know, it's going to give you fairly accurate targets.

00:22:34: But also because of the volume of targeting that is happening It's impossible for you verify.

00:22:40: how can a single person look and say out of this multi terabytes data?

00:22:44: This decision came up.

00:22:46: So what ends up?

00:22:47: Happening Is that you have Decision fatigue And you essentially default To being a rubber stamp or these AI systems.

00:22:53: so Even if you insert human in the loop, and these systems where you have so many decisions to make on a very short time.

00:23:01: It essentially becomes whatever is recommended just gets approved.

00:23:06: So I feel like we need to step back a little bit And digest this technology before We kind of over rely on it because it's A powerful technology but not perfect yet.

00:23:20: But do you

00:23:20: see... Like the school bombing, right?

00:23:25: The targeting of the missile to the school.

00:23:27: That was part of that recommended set of targets.

00:23:31: It's unclear why but I mean there may be wrong data There maybe misinformation Or it just made a mistake and hallucinated?

00:23:39: We don't know.

00:23:40: But we have in our time, we have the Saturday and Sunday.

00:23:44: We have two days off normally And I think something that we need too for the AI world.

00:23:50: but now it's always accelerating!

00:23:52: We can't catch up with all of news from China or the US sometimes from Europe.

00:24:00: So what could be done to stop acceleration?

00:24:06: At the end of day, I think.

00:24:08: The biggest leverage point is really government needs to get involved and unfortunately within the capability of technology, we're actually saying go faster.

00:24:26: I don't know what you saw but just a couple days ago they national policy framework for AI came out from US White House and it pretty much said that we are now going to allow data centers be built faster.

00:24:37: We will allow full copyrighted content inside AI training.

00:24:42: there's no liability And uh...we'll take away liability form these companies doing.

00:24:51: So it's everything that the Silicon Valley lobbyists have been asking for is now being put into regulation, whereas what we should be doing more is thinking about OK, should we slow down these a little bit?

00:25:08: Do we need to have another two thousand data centers in America which already has five thousand compared to the rest of the world, which is already?

00:25:17: you know it's like eighty percent of the World's data centers in America are ready.

00:25:20: Do we need more?

00:25:23: And how should this stuff be applied so that its actually being used for benefit of people versus just a few labs?

00:25:35: Should we be working together more?

00:25:36: I mean, We've talked about this in the past and you know.

00:25:40: This is where Europe can actually play a bigger role because Europe in general has had much more of a humanistic perspective In terms of how technologies relationship with people.

00:25:51: And if Europe can stand up a little bit help to think about what are the policies that allows this technology to grow safely, allow these technologies to be diffused in a way it does not create societal chaos.

00:26:06: I mean those kind of things creating better social safety net and maybe also in Europe as well.

00:26:12: Beacon light right now is very dark world.

00:26:18: there's role played versus being afraid.

00:26:23: Europe is falling behind, we actually need to do what America does.

00:26:29: I think that's the wrong narrative because America right now it going into a self-defeating, self harming race Because all of things that America doing try and get ahead will be exact reasons why they arrive at destination before its ready economic and societal chaos in its own society.

00:26:54: And maybe also create an AI that is harmful, less safe which endangers the entire

00:27:01: world.

00:27:02: Oh, wow.

00:27:03: That's a lot of stuff and but to stand up we have to be strong.

00:27:07: We learn to get stronger right now But it takes time.

00:27:10: who haven't learned it?

00:27:12: And I've never thought that i will ask for regulation in my life because I said regulation is very bad and We can do by ourself?

00:27:25: We need rules, we need something.

00:27:27: We need game changers in this and we need politician of the politicians field who is smart and under overview what's going on?

00:27:38: What can't over you right now?

00:27:41: because don't know exactly what will happen next three years.

00:27:45: Yeah, and it's actually probably gonna be faster right?

00:27:48: I don't know if you saw the article.

00:27:49: I sent about the mini max at two point seven.

00:27:52: So that's a new version of The Chinese minimax model That they essentially just had to go into a loop of self-improvement in within a few weeks of two point five coming out.

00:28:03: Two point seven was thirty percent better than then.

00:28:06: then then two point and there was nobody that actually went into the code.

00:28:11: It would just be AI improving itself, right?

00:28:13: And at this type of a rate if it then sends out to they make two point eight may two-point nine.

00:28:20: The speed of improvement is going to Be exactly what you said earlier because at a speed that we cannot keep up.

00:28:30: So when people go back and we're actually I want to emphasize on things that what People Go Back and they always say oh, but you look at the prior industrial revolution.

00:28:36: There was always more jobs created.

00:28:37: We Always Adapted.

00:28:38: we are very adaptable Species That's True.

00:28:43: When When?

00:28:45: Revolutions And Industrial Evolutions Come in Decades right and The First Second of Third Came In eighty, sixty and forty years of increments to play out.

00:28:55: We are now playing a full industrial revolution in three or four five years no matter how adaptive we as species.

00:29:04: our society needs time for that to play into the economy and into our society And slowing down is actually this smartest thing

00:29:17: it's for us.

00:29:17: It's nearly late in the afternoon, I need a beer you'll need to coffee

00:29:22: your early

00:29:23: in the morning because after all this stuff i need something positive and i think that's beer and sushi for me.

00:29:31: but Alvin thank you very much.

00:29:34: yeah we need orientation We need facts.

00:29:38: You are giving to us And I can only thank you.

00:29:43: Yeah, let's keep up and say what will happen in the future.

00:29:47: Let's talk about it!

00:29:50: I'm so glad that we have this ability to converse every two weeks overly framed on how do we go faster, and make more profits.

00:30:09: Whereas really what's needed right now is the real impact of technology?

00:30:15: And will society adapt to it in a non-chaotic way?

00:30:26: Value to humanity versus creating greater value.

00:30:28: two a few individuals.

00:30:29: I think that's something that we need to Elevate in the conversation,

00:30:34: you just slow down initiative

00:30:38: Yeah Which which is very counter?

00:30:40: I know too and then I'm from this Silicon Valley And they're very focused on going faster.

00:30:46: But this was one of the few times an animal.

00:30:48: I'm a technology optimist.

00:30:50: This is one of those few times in history where I really think that slowing down will actually benefit the world and even benefit people who are trying to accelerate because they think that there going make trillions of dollars by making AGI.

00:31:06: But what don't realize is if we create a GI before it's ready, then we'll create chaos which destroys economy.

00:31:18: There's nothing positive, Evan.

00:31:23: Please!

00:31:25: The one positive thing I want to say is that we still have time To affect the outcome and so having more people listen.

00:31:35: what were talking about?

00:31:36: Having more policy makers start taking action on some of these things actually some more regulation and policy on safety, and on you know soft landing policies.

00:31:49: On international cooperation.

00:31:50: if we can start working on these things?

00:31:52: We have probably somewhere between a one to two-year window where to change the speed and change trajectory.

00:32:01: So let's benefit from that one two years window, And once again Alvin thank you very much!

00:32:08: In two weeks we'll see each other then with a lot of positive news!

00:32:13: Very good I look forward too!

00:32:15: Okay Thank You!

00:32:16: Bye bye!

00:32:17: Have a great

00:32:17: day!