SMW '19: IBM的斯科特 Grizzle谈论人工智能和沃森的新进展

以下是采访全文:

蒂姆Siglin: 欢迎回到流媒体西部2019, if you hear the alarms in the background that's not because they're coming after me; it's because something's going on in the hotel. I've got with me today 斯科特 Grizzle from IBM 和斯科特 actually has been in the industry as long as I've been. 你说参加这个节目已经20年了，你的第一个节目是在哪里?

斯科特灰色: 第一场演出是在圣何塞.

蒂姆: 在麦克奈瑞会议中心.

史考特: 这是第一次，这是一场盛大的演出.

蒂姆: 确实如此.

史考特: And then we did other locations, then came back but it was up there for years.

蒂姆: And I remember in fact there was the Westin that was right across the street from that. 圣. 我想，它的名字叫克莱尔. 那是在他们把... 好吧 before they put the parking deck in they actually had a place where you could ice skate in the winter.

史考特: 这是正确的.

蒂姆: 就在下面. 这就是我们的年代. 讽刺的是我们现在又回到了洛杉矶市中心的威斯汀酒店.A. I honestly liked the having it up in Silicon Valley but I underst和 also why they've moved it south to be a part of the entertainment capital. 你在节目播出前提到的... 这周流媒体上发生了很多事.

史考特: 是的.

蒂姆: So there are multiple shows, sports, Esports, there were a couple other ones you mentioned.

史考特: Salesforce和Dreamforce就是这样.

蒂姆: 对，梦幻部队.

史考特: 那里正在发生一件大事. 我知道欧特克在做一些事情. 有无数不同的小事件在发生, 或者说不是很少, 这周有很多大事要办. 这是一个很合适的时间来做这个节目.

蒂姆: 它是, 你们知道, maybe it's that run-up to just before Thanksgiving where everything goes into a lull. So, 你在这个行业里转来转去, 你在IBM工作, I think you've been with IBM for the last two or three times that we did interviews with.

史考特: 差不多六年了. Ustream和IBM大约有6年的时间.

蒂姆: 哦，对了，因为你在Ustream. 那么你的工作重点是什么，你现在在IBM的头衔是什么?

史考特: So I just transitioned from being the lead sales engineer to channels 和 alliances. 我有了新的合作伙伴, 让他们注册, 和, 你知道, 确保我们能更有效地建立合作伙伴关系.

蒂姆: 好吧. 沃森是我们在过去的采访中谈到的事情之一. 此时视频的焦点在哪里?

史考特: 我们仍然做视频丰富，它可以做元数据提取, 场景分析, 成绩单, 理解别人在说什么, 做场景选择. 显然，封闭字幕非常大. 事实上，我们的标题还获得了读者选择奖的提名. 我们只是想把它们都做成一个统一的包装. 让它在我们的流媒体平台上播放. 下一个层次:使用人工智能来帮助增强不同方面.

蒂姆: 我想到了一件事... 我在今天早上第一次主题演讲时想的, 哪个是亚马逊那个拿着火棒的先生, 他说的是语音控制, 推出语音控制. Obviously with Watson 和 a lot of the underlying speech detection engines that IBM has, 他是在描述一个实际应用. What are some of the other applications from a consumer-facing st和point that AI is good for?

史考特: 好吧, if you look at how the cable companies are using it for the remote controls. 所以，你可以对它说话，你可以找到东西. 再一次。, 正确的语音转文本, 以及正确的搜索算法, 你可以做快速搜索和推荐. So I'm looking for, 你知道, Game of Thrones, 和 then here are other shows like Game of Thrones. 或者是你所要求的建议. 它可以寻找模式. 但如果你的孩子登录了, 他们看着小鲨鱼, 他们没有《百家乐软件》推荐.

蒂姆: 好吧, 和 that is one of the issues that we've had in the past is the recommendation engine was smart in terms of discovery, but not so smart in underst和ing that there are different users that are a part of it. So does it also look to sense that it's different voices as opposed to somebody actively having to log into different accounts, or...

史考特: 是的, 完全, 我的意思是 you can definitely see in the closed captioning you can do a notation for different speaker change. 男演讲者或女演讲者. 儿童演说家. 因为很明显，孩子的声音和成年人的声音大不相同. So you can do different things like that, 和 it'll show up faster in different aspects.

蒂姆: 那么是什么呢?, 以你现在所处的角色，联盟和伙伴关系, 你能举出一些实际的例子吗? 我的意思是, 很明显, 有些问题可能很棘手，你不能讨论, 但是从合作的角度来看，IBM想要做什么呢?

史考特: 有了人工智能, 所以我们有一些客户做得很好, 在很多情况下几乎是CMS. 所以他们将视频丰富与流媒体与其他api连接起来. 做翻译. 学习多种语言. So the whole idea is how you can interconnect, bring the metadata, 和 do the extraction. 因为我们有元数据，所以您可以轻松地将其引入，以便更快地进行搜索. 有了更多的信息，我们创造了新的用户体验. And the whole idea with our partners is how can they have the next-level offering that their competitors have. Differentiators between what they have 和 what somebody else has, 和 the usability. 因为如果有人能快速搜索, 还有一种强大的寻找方式, 而不是另一个视频平台. 你知道，就是一个仓库. 我的意思是, 我有一个客户，他一年举办大约600场活动, 现在正在尝试做UGC内容. 他们不知道自己的内容是什么. 怎么搜索呢? 你得做记录，做搜索引擎. 做算法. 所以现在它变得更快，就像他们有一个类固醇CMS. 所以现在你可以找到更个性化的东西.

蒂姆: 最后一个问题. 我和一些人一起工作过 ... 年前. 当然，你也曾与IBM沃森和语音转文本合作过. One of the tricks, or one of the challenges is always been multiple languages simultaneously. 所以我问了亚马逊的那位先生, 他说最终他们把所有的语言都装进了Alexa设备. 不是只有印地语、西班牙语或英语. But I was asking about the question of people who tend to go back 和 forth between languages when they're talking. 你们都面临过这些挑战吗? Or is it essentially a partner would license a single language as opposed to all languages?

史考特: 我们的很多合作伙伴都在做单一语言. 母语等等. 但是IBM也有翻译服务. 他们实际上有沃森api. 它们也有标准的多种语言. 所以从理论上讲，他们可以通过和改变多种语言和搜索. 但问题是，这些方面没有训练引擎. 所以这有点不同. 但在沃森传媒那边, our whole conference is making a user experience that is a little more trainable in one spot versus build your own trainer on top of what you have. 为了方便消费者.

蒂姆: 对，所以本质上你有更多的堆栈比你提供的.

史考特: 正确的.

蒂姆: 好吧, 好吧, 斯科特, as always it's always a pleasure to speak to you 和 we'll be right back here in just a few minutes with our next interview.