撰文 | 杨丽
在一份新报告中，研究机构 Nucleus Research 确定了今年的热门技术主题：AI，物联网，安全和供应链应用将在 2017 年变得更加集中且可实施。
根据 Nucleus Research 的最新一期预测，企业和 IT 部门曾相信能提供比公共云更安全的本地部署环境，但现在这类想法已不复存在。
事实上，一语成谶，该机构将这种衰落描述成为「可怕的死亡正在发生」。分析师团队在 10 月 19 日发布的「2017 年十大预测报告」中写道：「明年，几乎不会有哪个人会认为本地部署安全具有一定的优越性。」
这个榜单由 Nucleus 的 Seth Lippincott，Anne Moxie，Barbara Peck，Brent Kinner，Moira Smalley 和 Rebecca Wettemann 共同撰写。他们承认，机构分析师在去年做出了相同的预测，目前这个预测还没有成真。
然而，分析师写道，云提供商在云安全方面投入的精力远远超过了本地部署部门的预估。他们不提云服务供应商应如何设计一个更加统一的环境，并维护几款操作系统的标准副本，这与企业 IT 商店中典型的万能软件不同。
尽管如此，分析师们在 2017 年的结论中清楚地表明：客户数据「在本地部署的云端能得到更好的保护」。他们写道：「甚至是坚信本地部署的 IT 部门也会妥协，承认他们比不上云供应商提供的安全性。
就人工智能而言，供应商必须仔细审查，以确定他们是否真的切实可行。2016 年新技术的报告不断涌现，新的 AI 系统也变得触手可及，但能力的落实却远远跟不上这类学术报告的递进。
Nucleus 分析师指出，新版 HBO 电视剧《西方世界》，或老版电视科幻小说系列《人类》中出现的 AI 系统，代表着许多人现在所认为的 AI 系统未来所具有的能力。
「在实践中，AI 远远没有达到它的潜力，」他们警告。实际提供 AI 技术的供应商将它连接到机器学习和某种形式的人机界面，包括音频，视觉和自然语言。
谷歌 DeepMind 开发的人工智能系统 AlphaGo 围棋战胜世界冠军李世石，同时，IBM 的人工智能「沃森」在 Jeopardy 竞赛节目中击败了人类选手！这个胜利让我们看到了强大的 AI 系统可以走得更远。但在科幻小说和实际情景中，AI 技术却存在着「显著的差距」。
根据报告，「2017 年将是关于哪些供应商真正能够开发 AI 技术，并将其纳入到他们的应用程序的一年」。
Nucleus 分析师还预测，对于大型企业软件厂商，如 SAP 和 Oracle，其第三方咨询将在 2017 年成为更重要的因素，并扩展到辅助角色。尽管 Oracle 受到诉讼拖累，但第三方咨询公司，如 Rimini Street 和 TomorrowNow（2005 年被 SAP 收购）将继续发挥作用，并持续发展。
在另一个预测中，Nucleus 分析师表示 2017 年嵌入供应链的机器学习和人工智能系统将自动检测和纠正交付过程中的错误。从过去的错误中学习的算法将有助于纠正出现的错误。
以上是 Nucleus Research 对 2017 年五大 IT 技术排名预测。
Cloud Security, AI, IoT Make List Of Hot Technologies For 2017
By Charles Babcock
In a new report, Nucleus Research identifies which of this year's hot tech topics -- including AI, IoT, security, and supply chains applications -- will become even more focused and implementable in 2017.
Those enterprises and IT departments that believe they are maintaining a more secure environment on-premises than the public cloud provides are a dying breed, according to the latest round of predictions from Nucleus Research.
In fact, the research firm described the decline in more apocalyptic terms: A great die-off is occurring. "The last believers in the superiority of on-premises security will follow the dinosaurs into the annals of history" next year, its analyst team wrote in its "Top 10 Predictions for 2017" report, released Oct. 19. The list was compiled by Seth Lippincott, Anne Moxie, Barbara Peck, Brent Kinner, Moira Smalley, and Rebecca Wettemann at Nucleus. The firm's analysts made the same prediction last year, they acknowledge, and it hasn't come true yet.
Nevertheless, the superior investment in security made by cloud providers "far outstrips what on-premises systems could hope to deliver," the analysts wrote. They made no mention of how cloud service suppliers design a more uniform environment and maintain a few standard copies of operating systems, instead of having one of everything, as is often typical in an enterprise IT shop.
Still, the analysts are clear in their conclusion for 2017: customer data "is far better protected" in the cloud that is on-premises. "Even the most die-hard on-premises true believers will emerge from their caves and admit that they can't match the security offered by cloud vendors," they wrote.
If that was the leading conclusion on the Top 10 list, another was that, for all the talk about artificial intelligence, vendors must be carefully scrutinized to ascertain whether they've really got the goods. The year 2016 was about claims of advances and new AI systems becoming available, but claims may be exceeding capabilities.
The Nucleus analysts pointed to the AI systems in the new HBO TV series,Westworld, or the older TV science fiction series, Humans, as representing what many people now think AI systems are capable of, or will be soon.
"In practice, AI is far from reaching its potential," they warn. Vendors who actually offer AI will have it connected to machine learning and some form of human interface, whether audio, visual, or natural language.
Google's AI system won the game of Go in March, a parallel IBM's Watson beating human contestants in Jeopardy! The win gave a glimpse of how far powerful AI systems can go. But there's "still a significant gap" between portrayals in science fiction and AI's accomplishments in practical settings on the ground.
"2017 will be about which vendors are truly able to develop it and incorporate it into their applications," according to the report.
The Nucleus analysts also predicted that third-party support for major enterprise software, such SAP and Oracle, will become a more important factor in 2017 and expand into auxiliary roles. Despite being the target of suits by Oracle, third-party support firms such as Rimini Street andTomorrowNow (acquired by SAP in 2005) will continue to function and thrive.
In the future, such firms will not only provide support for a set of applications, but will also act as more of a partner to its customers, helping them find new sources of software and ways to adopt it. The third-party support tends to reduce the annual maintenance cost of legacy software by up to 50% over what the original supplier charges, the analysts wrote. Companies need that savings to develop their next project in the cloud.
While the internet of things has been in the forefront for over a year, the analysts find that 2017 will mark the year when it gets out of its "shiny object syndrome" and becomes another core enterprise system with an impact on the bottom line.
Companies in manufacturing, agriculture, and oil and gas have already implemented some form of IoT. The lessons learned from early implementations will make it more likely that IoT will creep into many other industries, they predicted. "Seeing IoT should be expected and should no longer be considered a novelty," the report finds.
In another prediction, the Nucleus analysts said 2017 will see machine learning and artificial intelligence systems embedded in supply chains to automatically detect and correct errors in the order to delivery process. Algorithms that learn from past mistakes will help correct emerging ones as they occur.
Previously, supply chain vendors sought to provide visibility into an end-to-end process, but they're now rushing to try to build in self-correcting measures that free the supply chain from dependence on a supply chain manager catching the error, once it's been made visible.
本文分享自微信公众号 - 人称T客（Java_simon），作者：T客汇
原文出处及转载信息见文内详细说明，如有侵权，请联系 firstname.lastname@example.org 删除。
【导读】继Google DeepMind在Nature上发表最新论文AlphaGo Zero的重磅消息，今天 AlphaGo 团队被国际人工智能会议（IJCAI...