1 00:00:00,120 --> 00:00:04,920 Speaker 1: Microsoft CEO Sacha and Adela who joins us now live Satcha. Obviously, 2 00:00:04,960 --> 00:00:09,920 Speaker 1: it's been a very fast moving and dramatic last seventy 3 00:00:09,920 --> 00:00:12,000 Speaker 1: two to forty eight hours, And I want to start 4 00:00:12,000 --> 00:00:14,800 Speaker 1: with the state of play. Where are we now? Open 5 00:00:14,840 --> 00:00:17,720 Speaker 1: ai says it's hired a new CEO. We're reporting their 6 00:00:17,800 --> 00:00:21,400 Speaker 1: efforts to get Sam Altman back. Tell us where things stand? 7 00:00:22,520 --> 00:00:24,960 Speaker 2: Well, first, for all, Havan, it's great to be with you. 8 00:00:24,960 --> 00:00:28,800 Speaker 2: You know, on Friday morning, we were very excited, committed 9 00:00:28,840 --> 00:00:32,560 Speaker 2: and confident in our innovation roadmap to bring this next 10 00:00:32,600 --> 00:00:37,360 Speaker 2: generation of AI to our customers. And we feel fantastic 11 00:00:37,479 --> 00:00:40,839 Speaker 2: and capable of doing that. And we were partnered with 12 00:00:40,880 --> 00:00:43,840 Speaker 2: open ai, and we were partnered with Sam. And that's 13 00:00:43,880 --> 00:00:46,920 Speaker 2: exactly where I am on Monday afternoon, if you will, 14 00:00:47,280 --> 00:00:49,519 Speaker 2: because we think that we can, you know, we are 15 00:00:49,600 --> 00:00:53,360 Speaker 2: leading in this next generation of AI technology. We continue 16 00:00:53,360 --> 00:00:55,440 Speaker 2: to be committed to open Ai, and we continue to 17 00:00:55,440 --> 00:00:58,720 Speaker 2: be committed to Sam and Greg and the team in 18 00:00:58,760 --> 00:01:00,840 Speaker 2: respect you where they are. And you know, I think 19 00:01:00,840 --> 00:01:04,759 Speaker 2: about Sam has chosen multiple times now to work with us, 20 00:01:04,920 --> 00:01:07,240 Speaker 2: and that's fantastic to see. And I think the real 21 00:01:07,319 --> 00:01:11,120 Speaker 2: thing is that the capability that Microsoft has across the 22 00:01:11,319 --> 00:01:15,520 Speaker 2: tech stack is what attracts great people like Sam, you know, 23 00:01:15,560 --> 00:01:18,080 Speaker 2: and people like Sam and you know, innovators like Sam 24 00:01:18,120 --> 00:01:20,240 Speaker 2: when it comes to ai to come to us, and 25 00:01:20,280 --> 00:01:21,080 Speaker 2: we are thrilled about it. 26 00:01:22,240 --> 00:01:28,160 Speaker 1: You incredibly quickly hired Sam as well as Greg. We 27 00:01:28,280 --> 00:01:32,360 Speaker 1: are hearing that Sam wants to return. Investors want him 28 00:01:32,560 --> 00:01:35,959 Speaker 1: to return to open Ai. How would you feel about that? 29 00:01:37,200 --> 00:01:40,520 Speaker 2: Yeah, as I said, we really want to partner with 30 00:01:40,560 --> 00:01:42,640 Speaker 2: open Ai and we want to partner with Sam. And 31 00:01:42,680 --> 00:01:45,640 Speaker 2: so in respect you where Sam is, He's working with 32 00:01:45,760 --> 00:01:48,680 Speaker 2: Microsoft and that is the case on Friday, and that'll 33 00:01:48,680 --> 00:01:51,840 Speaker 2: be that's the case today and we will I absolutely 34 00:01:51,840 --> 00:01:53,280 Speaker 2: believe that will be the case tomorrow. 35 00:01:54,400 --> 00:01:58,400 Speaker 1: So what are the conversations you've had with open AI's 36 00:01:58,560 --> 00:02:03,240 Speaker 1: current board from their perspective, where do things stand? And 37 00:02:03,280 --> 00:02:06,920 Speaker 1: have you talked with Emmett Sheer, the new interim CEO. 38 00:02:08,360 --> 00:02:10,959 Speaker 2: Yeah, I've had conversations with Emmett, and again it's the 39 00:02:11,000 --> 00:02:14,400 Speaker 2: same thing. There's no real difference there from where we 40 00:02:14,400 --> 00:02:16,840 Speaker 2: were when we were working with Mira and she was 41 00:02:16,880 --> 00:02:20,720 Speaker 2: interim CEO, when Sam was CEO and Emmett. So my 42 00:02:20,800 --> 00:02:23,120 Speaker 2: message to them, Emmett is very clear, which is, Hey, look, 43 00:02:23,160 --> 00:02:25,480 Speaker 2: we remain very very committed to open ai and it's 44 00:02:25,600 --> 00:02:28,440 Speaker 2: mission and it's sort of roadmap, and they can count 45 00:02:28,440 --> 00:02:31,360 Speaker 2: on us. And then, as I said, we also are 46 00:02:31,440 --> 00:02:33,600 Speaker 2: very committed to Sam and Greg and team that want 47 00:02:33,639 --> 00:02:36,200 Speaker 2: to join us. If they're not at open ai or 48 00:02:36,240 --> 00:02:38,959 Speaker 2: anyone else who's at open ai wants to go somewhere else, 49 00:02:38,960 --> 00:02:40,920 Speaker 2: we want them to come to Microsoft and continue to 50 00:02:40,960 --> 00:02:43,720 Speaker 2: work here and win partnership with open ai. 51 00:02:44,919 --> 00:02:48,720 Speaker 1: To your knowledge, why was Sam fired? And to your 52 00:02:48,840 --> 00:02:52,400 Speaker 1: knowledge was he involved in any wrongdoing? Has the board 53 00:02:52,520 --> 00:02:54,400 Speaker 1: given you a reason? 54 00:02:56,720 --> 00:02:59,040 Speaker 2: As far as I'm concerned, You know, we were, as 55 00:02:59,040 --> 00:03:01,920 Speaker 2: I said, we were very confident in Salmon his leadership team. 56 00:03:02,480 --> 00:03:05,359 Speaker 2: I've not been told about anything you know, they published 57 00:03:05,360 --> 00:03:09,360 Speaker 2: internally at open ai that there is not that the 58 00:03:09,400 --> 00:03:12,000 Speaker 2: board has not talked about anything that Sam did other 59 00:03:12,080 --> 00:03:14,880 Speaker 2: than some breakdown in communications. And you know, I'm not 60 00:03:14,919 --> 00:03:18,040 Speaker 2: directly and was told by anyone from their board about 61 00:03:18,080 --> 00:03:21,120 Speaker 2: any issues. And so therefore I remain confident in Sam 62 00:03:21,160 --> 00:03:24,720 Speaker 2: and his leadership and capability. And that's why you know, 63 00:03:24,800 --> 00:03:26,680 Speaker 2: we want to welcome him to Microsoft. 64 00:03:27,840 --> 00:03:31,280 Speaker 1: Now. We understand that to support a return of Sam 65 00:03:31,280 --> 00:03:34,920 Speaker 1: momon too open Ai. Microsoft wants some changes to the board, 66 00:03:35,000 --> 00:03:38,480 Speaker 1: to governance to its overall contract with open ai, so 67 00:03:38,600 --> 00:03:42,480 Speaker 1: something like this never happens. Again, what specifically are you 68 00:03:42,560 --> 00:03:45,560 Speaker 1: looking for, for example, would you want to board see 69 00:03:45,760 --> 00:03:46,840 Speaker 1: and if not, what else? 70 00:03:47,960 --> 00:03:49,640 Speaker 3: Yeah, I mean I think we definitely will want some 71 00:03:49,680 --> 00:03:50,560 Speaker 3: governance changes. 72 00:03:50,600 --> 00:03:53,640 Speaker 2: So then you know, you know, surprises are bad and 73 00:03:54,440 --> 00:03:56,800 Speaker 2: we just want to make sure that things are done 74 00:03:56,840 --> 00:03:58,520 Speaker 2: in a way that will allow. 75 00:03:58,360 --> 00:03:59,920 Speaker 3: Us to continue to partner. 76 00:04:00,080 --> 00:04:03,000 Speaker 2: Well that's about it, right, You know, this idea that somehow, 77 00:04:03,120 --> 00:04:06,040 Speaker 2: you know, suddenly changes happen without being you know, in 78 00:04:06,080 --> 00:04:09,640 Speaker 2: the loop is not good. And we will definitely ensure 79 00:04:09,680 --> 00:04:11,680 Speaker 2: that some of the changes that are needed to happen 80 00:04:11,920 --> 00:04:15,080 Speaker 2: and we continue to be able to go along on 81 00:04:15,120 --> 00:04:16,680 Speaker 2: the partnership with open Ai. 82 00:04:17,720 --> 00:04:20,560 Speaker 1: So how are you envisioning this role with the sort 83 00:04:20,600 --> 00:04:23,000 Speaker 1: of you know, advanced AI team that Sam and Greg 84 00:04:23,000 --> 00:04:26,240 Speaker 1: would be joining and leading. Can you explain that? And 85 00:04:26,320 --> 00:04:30,120 Speaker 1: are they actually Microsoft employees right now? Like who do 86 00:04:30,160 --> 00:04:30,720 Speaker 1: they work for? 87 00:04:32,000 --> 00:04:32,279 Speaker 3: Yeah? 88 00:04:32,279 --> 00:04:35,280 Speaker 2: So they're all in the process of joining, and yes, 89 00:04:35,360 --> 00:04:37,919 Speaker 2: I mean the thing is, you know, we have a 90 00:04:38,040 --> 00:04:40,880 Speaker 2: ton of AI expertise in this company. You know, in fact, 91 00:04:40,920 --> 00:04:44,159 Speaker 2: at our IGNITE conference last week. We talked about all 92 00:04:44,200 --> 00:04:47,200 Speaker 2: of the great work open aiyes doing and on Azure 93 00:04:47,320 --> 00:04:50,840 Speaker 2: and all the models and tools that we built around it. 94 00:04:51,080 --> 00:04:53,279 Speaker 2: And we also talked about all the open source models 95 00:04:53,320 --> 00:04:56,240 Speaker 2: that are on Azure from you know, Llama and miss 96 00:04:56,279 --> 00:05:01,240 Speaker 2: Stral and Jazz and everything you and then also you know, 97 00:05:01,279 --> 00:05:04,479 Speaker 2: in fact, not only are we leaders in LLLMS in 98 00:05:04,560 --> 00:05:07,599 Speaker 2: partnership with open Ai, but we're also leaders in SLMs 99 00:05:07,680 --> 00:05:09,960 Speaker 2: or the small language models with five and we talked 100 00:05:10,000 --> 00:05:12,320 Speaker 2: about that, which is all developed by Microsoft Research. So 101 00:05:13,080 --> 00:05:15,520 Speaker 2: what we am excited about is sort of creating another 102 00:05:15,600 --> 00:05:18,520 Speaker 2: team that's really going to have high ambition on leading 103 00:05:18,600 --> 00:05:21,160 Speaker 2: edge advanced day I work that Sam and Greg are 104 00:05:21,160 --> 00:05:23,839 Speaker 2: excited about. This is something that you know, we've talked 105 00:05:23,839 --> 00:05:26,080 Speaker 2: a lot about with them. You know, what happens in 106 00:05:26,120 --> 00:05:28,839 Speaker 2: twenty five, what happens in thirty What do we do 107 00:05:29,000 --> 00:05:32,040 Speaker 2: to advance both the system side the science side. 108 00:05:32,920 --> 00:05:34,920 Speaker 3: And that's the thing that we will be pushing on. 109 00:05:35,960 --> 00:05:38,680 Speaker 1: Now. Many folks I'm speaking with don't see Sam as 110 00:05:38,720 --> 00:05:41,840 Speaker 1: a sort of big tech company guy. He has all 111 00:05:41,839 --> 00:05:45,000 Speaker 1: of these side product projects that you know, we've already 112 00:05:45,040 --> 00:05:47,359 Speaker 1: heard about. We've reported on new projects. Would he be 113 00:05:47,400 --> 00:05:53,159 Speaker 1: able to pursue these side projects while being employed by Microsoft? 114 00:05:53,720 --> 00:05:56,960 Speaker 2: Yeah, I'm sure you know, Like you know, Sam has 115 00:05:57,000 --> 00:06:00,800 Speaker 2: got broad interests in broad investments and so and we'll definitely, 116 00:06:00,839 --> 00:06:03,400 Speaker 2: you know, work through the governance aspects of it. But 117 00:06:03,440 --> 00:06:05,960 Speaker 2: most importantly, I think you would want to only work 118 00:06:05,960 --> 00:06:08,640 Speaker 2: at Microsoft if he wants to spend his full time 119 00:06:08,720 --> 00:06:11,520 Speaker 2: time on really pursuing the mission, just like how he 120 00:06:11,560 --> 00:06:13,560 Speaker 2: were spending his time on Open AI. And so that's 121 00:06:13,640 --> 00:06:17,680 Speaker 2: kind of the idea behind having him lead this advanced 122 00:06:18,200 --> 00:06:21,160 Speaker 2: research team around AI with Greg and others. 123 00:06:21,520 --> 00:06:23,760 Speaker 3: And you know, that's an exciting thing for us to 124 00:06:23,800 --> 00:06:24,359 Speaker 3: look forward to. 125 00:06:25,880 --> 00:06:29,680 Speaker 1: I'm curious just your reaction to you know, the last 126 00:06:29,920 --> 00:06:33,320 Speaker 1: three days of events. Are you now more concerned than 127 00:06:33,360 --> 00:06:38,240 Speaker 1: ever about AI safety? Given we've seen clearly how fragile 128 00:06:38,520 --> 00:06:39,720 Speaker 1: these institutions are. 129 00:06:41,400 --> 00:06:42,240 Speaker 3: It's a great point. 130 00:06:42,279 --> 00:06:45,000 Speaker 2: I mean, the thing that I think, in some sense, 131 00:06:45,560 --> 00:06:49,200 Speaker 2: I've always felt that we should think about the unintended 132 00:06:49,240 --> 00:06:53,159 Speaker 2: consequences of any great advances in technology from day one, 133 00:06:53,360 --> 00:06:56,080 Speaker 2: versus dealing with them later. So in some sense I 134 00:06:56,279 --> 00:07:01,800 Speaker 2: welcome this dialogue, if you will, of safe and safety first, 135 00:07:01,920 --> 00:07:05,080 Speaker 2: even in technology and technological development. 136 00:07:05,160 --> 00:07:05,240 Speaker 3: Right. 137 00:07:05,279 --> 00:07:07,320 Speaker 2: I mean that's I think in some sense it should 138 00:07:07,360 --> 00:07:10,160 Speaker 2: be celebrated that we as a tech industry have gotten 139 00:07:10,240 --> 00:07:12,840 Speaker 2: to a place where we don't think of this as 140 00:07:12,880 --> 00:07:15,120 Speaker 2: a trade off, but we think about this as two 141 00:07:15,160 --> 00:07:17,880 Speaker 2: considerations that are both first class. Right. One is how 142 00:07:17,920 --> 00:07:21,160 Speaker 2: do we build technology so that everybody in the world 143 00:07:21,200 --> 00:07:24,920 Speaker 2: can have a doctor an X that they can reach, 144 00:07:25,040 --> 00:07:28,280 Speaker 2: or a tutor or you know, a rural farmer in 145 00:07:28,280 --> 00:07:31,480 Speaker 2: India has more agency because of technology that was developed 146 00:07:31,520 --> 00:07:33,160 Speaker 2: in the West Coast of the United States a few 147 00:07:33,200 --> 00:07:37,000 Speaker 2: months earlier. These are unbelievable things that democratize access on 148 00:07:37,000 --> 00:07:39,600 Speaker 2: one side, but on the other side, being grounded on 149 00:07:39,680 --> 00:07:43,200 Speaker 2: the here and now harms right, whether it's election interference 150 00:07:43,360 --> 00:07:47,520 Speaker 2: or deep fakes or bias or bioterrorism and having the 151 00:07:47,520 --> 00:07:50,040 Speaker 2: guardrails against it or even some of the you know, 152 00:07:50,080 --> 00:07:53,840 Speaker 2: AI take off existential risks and the alignment research that 153 00:07:53,920 --> 00:07:55,920 Speaker 2: is required. So I think that there is a real 154 00:07:56,080 --> 00:08:00,480 Speaker 2: robust dialogue that's happening, which is and real work. Like 155 00:08:00,520 --> 00:08:02,880 Speaker 2: when I think about Microsoft, we have done, in fact, 156 00:08:02,920 --> 00:08:06,520 Speaker 2: some of the biggest most work around AI alignment, AI 157 00:08:06,640 --> 00:08:09,920 Speaker 2: safety guardrail. So for example, even the open source models 158 00:08:09,960 --> 00:08:13,200 Speaker 2: we launched last week all have benefit from all the 159 00:08:13,200 --> 00:08:16,800 Speaker 2: godrail work we did around open ai models. 160 00:08:17,920 --> 00:08:20,720 Speaker 1: So throughout the reporting process, we've heard, you know, concerns 161 00:08:20,720 --> 00:08:23,440 Speaker 1: about potential regulatory issues and of course the power that 162 00:08:23,480 --> 00:08:26,800 Speaker 1: Microsoft already has, the power that Microsoft has in Ai. 163 00:08:26,880 --> 00:08:28,720 Speaker 1: Have you, on your end work through all the potential 164 00:08:28,800 --> 00:08:31,080 Speaker 1: legal issues at play here? I mean I just spoke 165 00:08:31,080 --> 00:08:34,000 Speaker 1: to one investor who said, you know, they're considering suing 166 00:08:34,080 --> 00:08:37,760 Speaker 1: open Ai. Like, you know, there's a lot of things 167 00:08:37,760 --> 00:08:40,400 Speaker 1: that have yet it seems to be figured out. And 168 00:08:40,440 --> 00:08:44,080 Speaker 1: do you know what Microsoft's legal liability is and open 169 00:08:44,120 --> 00:08:47,640 Speaker 1: AI's legal liability is given your significant investment there. 170 00:08:48,840 --> 00:08:53,920 Speaker 2: Yeah, Look, I'm I'm most focused on and most excited 171 00:08:53,960 --> 00:08:54,640 Speaker 2: about our. 172 00:08:54,520 --> 00:08:57,760 Speaker 3: Ability to continue to innovate Emily. 173 00:08:57,920 --> 00:09:01,720 Speaker 2: I mean, I led the lawyers sort of figured out 174 00:09:01,760 --> 00:09:05,359 Speaker 2: what liabilities are. But I think we have all the capability, 175 00:09:05,480 --> 00:09:08,680 Speaker 2: all the IP and all the things that we need 176 00:09:08,760 --> 00:09:12,400 Speaker 2: in order to continue to essentially control our destiny here 177 00:09:12,720 --> 00:09:14,440 Speaker 2: so that we can continue. 178 00:09:14,000 --> 00:09:16,079 Speaker 3: To innovate on behalf of our customers. 179 00:09:16,120 --> 00:09:19,920 Speaker 2: That's what matters, That's what makes Microsoft Microsoft, and will 180 00:09:19,920 --> 00:09:21,559 Speaker 2: continue to do what we need to do. 181 00:09:22,320 --> 00:09:25,600 Speaker 1: Okay, so quick, last question this is today, Who's going 182 00:09:25,640 --> 00:09:28,079 Speaker 1: to be CEO of open Ai tomorrow. You tell me. 183 00:09:30,080 --> 00:09:32,320 Speaker 3: I will leave it to open Ai and it's bored, 184 00:09:33,600 --> 00:09:34,040 Speaker 3: all right. 185 00:09:34,080 --> 00:09:35,600 Speaker 1: Sacha Adella, CEO of Microsoft. 186 00:09:35,640 --> 00:09:35,920 Speaker 2: Thank you. 187 00:09:35,960 --> 00:09:38,000 Speaker 1: I know you've had a really busy few days. Appreciate 188 00:09:38,000 --> 00:09:40,079 Speaker 1: you taking the time to talk to us.