As a mathematician, I use xAI's Grok regularly, as I find it vastly superior to the other models in terms of reasoning, logic etc. It could be argued that these are the key traits that must be pursued in order to reach AGI. And if reaching AGI is indeed a pivotal winner-take-all moment, then Grok's current little market share may not be a relevant factor (if the market share is even correct; many might use Grok via X – I don't know if it's in the statistics).
one important difference between hassabis and amodei is that hassabis runs deep mind, which is an alphabet subsidiary, while amodei runs anthropic, which is independent.
this does cut both ways. hassabis is at least nominally beholden to alphabet shareholders, and for things to go well in a deep mind-owned future, one has to trust not just hassabis and deep mind but also alphabet and its executives. on the other hand, while anthropic's independence and ai safety culture is presumably less likely to get co-opted by corporate influence, as a small company with less experience working with government, they may be more likely to get nationalized in a way that goes poorly.
i do think company culture below the executive level is also important, because beyond a certain size of organization the executive team only has so much influence on the workings of the company. historically anthropic has been the standout here (noting that i'm referring to regard for long-term risks from AGI here, not necessarily the geopolitical issues that you're primarily concerned with in this piece), though a friend who works there tells me that the culture has been diluted somewhat over the past few years as anthropic has become an attractive resume line for careerists.
As a mathematician, I use xAI's Grok regularly, as I find it vastly superior to the other models in terms of reasoning, logic etc. It could be argued that these are the key traits that must be pursued in order to reach AGI. And if reaching AGI is indeed a pivotal winner-take-all moment, then Grok's current little market share may not be a relevant factor (if the market share is even correct; many might use Grok via X – I don't know if it's in the statistics).
one important difference between hassabis and amodei is that hassabis runs deep mind, which is an alphabet subsidiary, while amodei runs anthropic, which is independent.
this does cut both ways. hassabis is at least nominally beholden to alphabet shareholders, and for things to go well in a deep mind-owned future, one has to trust not just hassabis and deep mind but also alphabet and its executives. on the other hand, while anthropic's independence and ai safety culture is presumably less likely to get co-opted by corporate influence, as a small company with less experience working with government, they may be more likely to get nationalized in a way that goes poorly.
i do think company culture below the executive level is also important, because beyond a certain size of organization the executive team only has so much influence on the workings of the company. historically anthropic has been the standout here (noting that i'm referring to regard for long-term risks from AGI here, not necessarily the geopolitical issues that you're primarily concerned with in this piece), though a friend who works there tells me that the culture has been diluted somewhat over the past few years as anthropic has become an attractive resume line for careerists.
AI reclines languorously on the chaise longue, while Robert Wright feeds it peeled grapes.
Thanks for your analysis of the AI players and of the coming threshold. I rely on NonZero to educate and keep me informed about AI.