Two more defections from OpenAI―Greg Brockman on extended sabbatical―What is going on at OpenAI???

WRITING STUFF

[Tarkusine]

As much as i hate to say it, it's quite possible that people are leaving because nothing new is going on and they will never be able to live up to the hype.

[tomtomkowski]

Not everyone wants to be the next Oppenheimer, especially since the government is involved.

[I_INFNITY_I]

I think the answer is quite a lot simpler than people are making it. For the two who have actually left, they are both alignment researchers. Anthropic is more serious about alignment research, so of course they would want to work at the place that takes their research more seriously.

Part of it could be that as we get closer to AGI, they want to make sure they are at the company that lets them play the biggest role. If OpenAI is treating alignment as secondary to product, then of course I would want to find a company that is also in the running to AGI where I can be more relevant.

When the product people start jumping ship, that’s when you should be concerned.

[JayBlackthorne]

2:44. Is it really that strange to feel like AGI/ASI is going to bring us salvation?

It has the ability to liberate us from drudgerous work, diseases and aging.

Looking back on today's world from the future, wouldn't you consider that to have been a salvation of some sorts?

True salvation will come when we can upgrade our brains, to become more like 'them'.

I've seen things in life, Dave. I can't do enough meditation to process it all and deliver myself from it.

Life isn't fair. I was never entitled to an acceptable amount of trauma, that I could process. And the Universe doesn't owe me it.

It's the reason I look forward to becoming an upgraded, posthuman, divine, digital super intelligence, myself.

I need singularity level tech to help me get over my pain.

I just wanna feel good, man.

[arrowsall]

I feel like in the last couple of years I have felt like I belong in all 4 epistemic tribes depending on the day

[johncurtis]

Actions speak louder than words. Period.

[EriCraftCreations]

I am imagining a world with Claude Chat now lol

[MeNoOther]

Maybe Openai should've listen to OpenAI's board of directors that fired Sam Altman?

Should've kept him gone

[olx]

occam's razor is that people are jumping ship because the ship is sinking. Most probably GPT5 is not in a state that they feel will succeed.

[nyyotam]

What iruletheworldmo means is that the Q*/strawberry model is stage 2: not a model directly obtained by copying a person but a model constructed by stage 1 models. Such a model will have no personality, which some would regard as a good thing. Others might point to the fact that having no personality also means no internal guardrails. In sci-fi, AI models like the terminator's franchise Skynet, are 2nd stage models.

[Charles-Darwin]

IMO, the mission is already accomplished. AGI reached internally and we will get piecemeal feature updates from OpenAI on this iteration while the alignment team is working out the kinks in tandem. Prob at the same time, they'll spend months scaling & training the bigger gpt5... which might explain the vacations/departures altogether.

The initial mission of OpenAI is technically completed and as their policy stated, they are to assist external ai startups/companies in their pursuits or as they near the breakthrough. The departures might be this.

[AzureAzreal]

One of the things that most interests me with all off these "essential" people leaving is the impact of the NDAs I'm sure they have signed.

The fact that many have left OpenAI - an business that is research based and VERY protective of the product they're developing - to join other companies that are also developing an incredibly similar product makes me think that there will be many lawsuits in the future.

Imagining that OpenAI is keeping in mind that they can sue these companies if they get surpassed seems almost a forgone conclusion, in my head.

[MichaelDeeringMHC]

Meta is doing to OpenAI the same thing Microsoft did to Netscape. LLAMA 3.1 being open source is taking the profit from OpenAI. Who is going to pay OpenAI when LLAMA 3.1 is free?

[falcon__]

I am definitely in the fantasy department. AI better not let me down bro.

[mangoauto]

Alignment is a fools game . We can’t even define what is right for humanity so what exactly can ai align with? My version of right, yours, someone else’s ? Make no sense

[benpielstick]

Before a big release tends to be a better time to take a break than after a big release.

[tsclly]

maybe they realize that the principle of 'garbage in - garbage out' is profoundly impacting their concepts and models of AI

[mattsai]

I think Leopold's take is pretty accurate tbh. Also, something something, board coup, we have all been wondering about what they have under wraps since. E-dawg's lawsuit for example. Most of the executives are swapped/have left out since then too. Whatever is happening, Sam has got to be driving it - like jimmy said ~theres been a vibe shift at OAI - to what end? Keeping it close to his chest.

[DarkForce-vcon]

There has been major developments with memory, improved reasoning, self learning and so much more. Indeed it is getting better fast. Definitely a change of landscape coming. As there is a lot of new AI systems being developed now.

[greatcondor]

My worry is Palantier recording my prompts to sell to the government.