If you search “AI is…” into Google, you end up with sayings from brilliant people like this:

如果您在Google中搜索“ AI is…”，那么最终您会得到像这样的才华横溢的人的话：

AI is the new electricity. — Andrew Ng, Coursera Cofounder

人工智能是新的力量。 — Coursera联合创始人 Ng

AI is more important than fire and electricity. — Sundar Pichai, CEO of Google

人工智能比火和电更重要。 — Google首席执行官 Sundar Pichai

AI is like teenage sex, Everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it.” — Dan Ariel, Duke Professor.

AI就像是青少年的性行为，每个人都在谈论它，没有人真正知道如何做，每个人都认为其他人都在做它，所以每个人都声称自己在做。” — 杜克大学教授 Dan Ariel 。

Sex, fire, and electricity. Oh my!

性，火和电。 天啊！

While it’s still early days, AI has proven to be more than just a fantasy. According to Adobe’s latest Digital Intelligence Briefing, top-performing companies are twice as likely to leverage AI technologies in their marketing than non-top-performers.

虽然还处于初期，但事实证明，AI不仅仅是幻想。 根据Adobe最新的Digital Intelligence Briefing，表现最佳的公司在市场营销中利用AI技术的可能性是非表现最佳的公司的两倍 。

Let’s put aside the analogies for AI for a moment and talk about what AI is, without the added rocket fuel. According to MIT Technology Review, AI, or Artificial Intelligence, can be defined as “machines being able to learn, reason, and act for themselves.” If you’ve noticed how your iPhone or Gmail predicts the next word you’re about to type, that’s AI.

让我们暂时搁置AI的类比，讨论在没有增加火箭燃料的情况下AI是什么。 根据麻省理工学院技术评论 ，人工智能或人工智能可以定义为“能够自我学习，推理和行动的机器”。 如果您注意到iPhone或Gmail会如何预测您要输入的下一个单词，那就是AI。

Despite the jargon, AI isn’t complicated to understand. A computer is fed data. An AI model (or algorithm) finds patterns within that data. The model provides a best guess on future outcomes based on patterns found in the data. As those guesses are tested out in the wild, the model considers new inputs to make better guesses next time.

尽管有术语，但AI并不复杂。 计算机被送入数据。 AI模型(或算法)在该数据内找到模式。 该模型根据数据中的模式提供对未来结果的最佳猜测。 当这些猜测在野外进行测试时，该模型将考虑新的输入以在下一次做出更好的猜测。

Today, AI has proven valuable in solving clear, defined problems. AI’s ability to process large amounts of data and generate actionable predictions in real-time highlights its advantage over humans. As marketers, we are perfect culprits for AI technologies — with no shortage of data, representing everything from behaviors like scroll data, numbers like web visits, images like advertisements, and even human emotions like happiness.

如今，事实证明，人工智能对于解决清晰明确的问题非常有价值。 人工智能能够处理大量数据并实时生成可行的预测，凸显了它相对于人类的优势。 作为营销人员，我们是AI技术的完美元凶-数据紧缺，代表一切，包括滚动数据之类的行为，网络访问之类的数字，广告之类的图像，甚至是幸福之类的人类情感。

Contrary to flashy headlines, AI does have limitations. AI models don’t understand the context the way humans do and, therefore, need to be instructed and exposed to what to pay attention to. “What makes human intelligence special is its adaptability — its power to generalize to never-seen-before situations,” says François Chollet, a well-known and respected AI engineer. AI still can’t do that though scientists are working on it.

与华而不实的头条新闻相反，人工智能确实有局限性。 人工智能模型不了解人类的工作方式，因此需要接受指导并暴露于关注的内容。 “使人类智能与众不同的是它的适应性-它具有将能力推广到前所未有的情况的能力，”著名且受人尊敬的AI工程师FrançoisChollet说。 尽管科学家正在努力，但AI仍然无法做到这一点。

The chances are high that AI won’t replace the need for creativity, but it will propel the speed in which we understand people’s behaviors, allowing us to deliver more personalized brand interactions.

人工智能不会取代创造力的可能性很高，但它将推动我们了解人们行为的速度，从而使我们能够进行更多个性化的品牌互动。

Let’s take a more in-depth look at where AI has shown the most promise across the marketing funnel and for whom, as well as areas where there’s still progress to be made.

让我们更深入地研究AI在整个营销渠道中显示出最大希望的领域以及针对谁，以及还有待发展的领域。

确定目标对象 (Identifying who to target)

Machine learning can quickly and proactively identify the best fit individuals or companies for you to market your product(s) to, an area where traditional lead scoring has fallen short.

机器学习可以快速，主动地找到最适合您的个人或公司，以供您将产品推销到传统潜在客户评分不足的领域。

Before diving into how AI can help, let’s take a quick look at the limitations of traditional lead scoring. Traditional lead scoring requires humans to assign a relative importance to each activity. This information, by nature, is subjective. The result is attributing value in a 1:1 relationship towards easily measurable actions, such as downloads or purchases.

在深入探讨AI如何提供帮助之前，让我们快速了解一下传统线索评分的局限性。 传统的铅评分要求人类为每种活动分配相对的重要性。 这些信息本质上是主观的。 结果是将1：1关系中的价值归因于易于测量的动作，例如下载或购买。

AI has proven advantageous in taking account not only the behavioral triggers on a company’s website but also external data points, like what types of websites someone visited or products he/she looked at.

事实证明，人工智能不仅可以考虑公司网站上的行为触发因素，而且还可以考虑外部数据点(例如，用户访问过的网站类型或他/她查看的产品)的优势。

In addition, by looking at engagement data on current customers, AI models can find similar behavior patterns to predict potential in-market buyers more accurately. As humans. we’ve likely glossed over or can’t recognize a connection for this data.

此外，通过查看当前客户的参与度数据，AI模型可以找到相似的行为模式，以更准确地预测潜在的市场购买者。 作为人类。 我们可能掩盖了该数据或无法识别该数据的连接。

B2B marketing platforms like 6sense have already invested in gathering those external intent signals and are using AI to help customers identify the best possible individuals and companies to market to, and giving them the tools to do so.

像6sense这样的B2B营销平台已经投资收集了这些外部意图信号，并正在使用AI帮助客户确定要推向市场的最佳个人和公司，并为他们提供实现这一目标的工具。

We know presenting your product or service at the time of need is critical. AI can not only help identify those people but act against it. Either arming the sales organization with information, presenting compelling digital advertising, or sending an email at opportune moments.

我们知道在需要时展示您的产品或服务至关重要。 人工智能不仅可以帮助识别这些人，而且还可以采取行动。 为销售组织提供信息，展示引人注目的数字广告或在适当时机发送电子邮件。

At this point, nearly every type of B2B and B2C company can benefit from advancements in AI and targeting.

在这一点上，几乎每种类型的B2B和B2C公司都可以从AI和定位方面的进步中受益。

确定要卖什么 (Identifying what will sell)

What if you could get in the minds of buyers before promoting (or even creating) new products? That’s the promise of a new field of AI & marketing known as neuromarketing and Emotional AI.

w ^帽子，如果你能得到购房者的头脑中推广(甚至创建)新产品之前？ 这是AI和营销新领域的承诺，即神经营销和情感AI。

As Harvard Business Review loosely defines it, Neuromarketing refers to the process of looking at “the measurement of physiological and neural signals to gain insight into customers’ motivations, preferences, and decisions.” It can inform product decisions as well as product attributes like price or advertising messages.

正如《 哈佛商业评论》所定义的那样，神经营销指的是“对生理和神经信号的测量以深入了解客户的动机，偏好和决定”的过程。 它可以告知产品决策以及产品属性，例如价格或广告信息。

Neuromarketing ranges from brain scanning to facial expression changes, to eye tracking movements. Today, most of the companies testing this technology are the goliaths — the likes of Google, Facebook, and Coca-cola. In certain scenarios, they find that what people self report vs. what is observed from brain scans differs.

神经营销的范围从大脑扫描到面部表情变化，再到眼动追踪。 如今，大多数测试此技术的公司都是老手，例如Google，Facebook和可口可乐。 在某些情况下，他们发现人们的自我报告与脑部扫描所观察到的有所不同。

Emotional AI start-ups take a slightly different approach to understanding sentiment, combining text, voice, and facial expression analysis. Like neuromarketing, the most common use is in pre-testing, replacing traditional consumer surveys to better understand what ad message or product features have the most positive impact on various groups of people.

情感AI初创企业采用略微不同的方法来理解情绪，将文本，语音和面部表情分析相结合。 与神经营销一样，最常见的用途是进行预测试，以取代传统的消费者调查，以便更好地了解哪些广告消息或产品功能对不同人群有最积极的影响。

The power of this information isn’t in understanding the sentiment, but making it actionable. That’s the goal of one start-up, Affectiva, who is trying to help companies interpret and respond to feelings at the moment — if someone is frustrated, do you send them a coupon code? Signal to the sales team to check-in on them?

这些信息的作用不在于理解情绪，而是使其可付诸实践。 这是一家初创公司Affectiva的目标，他正在尝试帮助公司解释和回应当前的感受-如果有人感到沮丧，您是否会向他们发送优惠券代码？ 向销售团队发出信号以签到？

As exciting as this sounds, according to VentureBeat, the technology isn’t accurate enough to interpret emotions yet. Humans are still better at reading and interpreting emotions than computers. And cultural variations in expression can make this information inaccurate or misleading. Despite that, it’s a $20 billion industry.

据VentureBeat称 ，这听起来令人兴奋，但该技术还不足以解释情绪。 与计算机相比，人类在阅读和解释情感方面仍然更好。 表达的文化差异可能会使此信息不准确或产生误导。 尽管如此，这还是一个价值200亿美元的产业。

A less expensive and invasive way to use AI is to analyze the web to understand broad level trends. Understanding trends can be achieved through data mining. For example, a company could analyze data across customer feedback channels to see what features are top of mind for consumers or lacking to help decide which features to highlight in marketing content.

使用AI的一种较便宜且具有侵入性的方法是分析Web以了解广泛的趋势。 通过数据挖掘可以了解趋势。 例如，一家公司可以跨客户反馈渠道分析数据，以了解哪些功能是消费者最关注的功能，或者缺少哪些功能可以帮助您决定在营销内容中突出显示哪些功能。

Ben & Jerry’s used machine learning to discover a new product marketing opportunity by analyzing cultural trends. Using AI models built on unstructured data, they found “at least 50 songs” that mentioned ‘ice cream for breakfast.’” The result led to a successful launch of a new line of cereal- flavored ice cream.

Ben＆Jerry的机器学习通过分析文化趋势来发现新产品营销机会。 他们使用基于非结构化数据的AI模型，发现了“至少50首歌曲”中提到了“早餐冰淇淋”。结果导致了一系列谷物味冰淇淋的成功推出。

确定推荐什么 (Identifying what to recommend)

One of the most common uses of machine learning today is in making smart recommendations. That can include product, content, or improve search capabilities.

如今，机器学习最常见的用途之一就是提出明智的建议。 这可以包括产品，内容或改善搜索功能。

Chances are, in the last day, you benefited from the output of a quality recommendation engine. It’s what you see on Netflix under “recommendations for you” under “recommended products” when you visit Amazon, under “made for you” on Spotify, and who you see late at night when you’re scrolling through Tinder.

在最后一天，您可能会从质量推荐引擎的输出中受益。 这就是您访问Amazon时，在Netflix上，在“推荐产品”下，在“推荐产品”下，在Spotify上的“为您量身定制”下以及在深夜滚动Tinder时看到的人。

However, these recommendation engines don’t all work the same way. The difference is in what data is fed into the algorithm. Some recommendation engines use what’s called “collaborative filtering.” That means if I like Lil Wayne, Kendrick Lamar, and Drake’s music. And you like Kendrick Lamar and Lil Wayne, that you might also like Drake’s music.

但是，这些推荐引擎的工作方式并不完全相同。 区别在于将哪些数据馈送到算法中。 一些推荐引擎使用所谓的“协作过滤”。 这意味着如果我喜欢Lil Wayne，Kendrick Lamar和Drake的音乐。 您喜欢Kendrick Lamar和Lil Wayne，也可能喜欢Drake的音乐。

Other recommendation engines layer in demographics, past behavioral attributes, and/or content or product similarities. To create a usable model, data that is readable by a computer must be provided. That can include text or machine vision to tag images.

其他推荐引擎位于人口统计，过去的行为属性和/或内容或产品相似性中。 要创建可用模型，必须提供计算机可读的数据。 这可以包括文本或机器视觉来标记图像。

Although I only mentioned B2C companies, recommendation engines can be just as useful for B2B companies with large amounts of content. By improving the conversion along the buyer journey, there’s a significant upside.

尽管我只提到了B2C公司，但是推荐引擎对于内容丰富的B2B公司同样有用。 通过改善购买者旅程中的转化率，存在很大的上行空间。

One of the reasons for the popularity in recommendation engines is it’s easy to track the value-add. According to Martech Today, the ROI can be measured in retention, profit, and engagement.

推荐引擎中受欢迎的原因之一是易于跟踪增值。 根据今天的Martech，可通过保留率，利润和参与度来衡量ROI。

The greatest risk with recommendation engines lies in social media and journalism, where feeding people more and more what they engage with can close off a whole world of ideas. For example, Tik Tok received negative criticism recently when an article suggested that they use image classification (race) to decide who and what to show you. If you follow young, blonde-haired, and blue-eyed women, you will be exposed to the type of content.

推荐引擎的最大风险在于社交媒体和新闻业，在社交媒体和新闻业中养活越来越多的人们，他们可能会封闭整个思想世界。 例如， Tik Tok最近受到负面批评，当时一篇文章建议他们使用图像分类(种族)来决定向谁展示什么。 如果您跟随年轻，金发和蓝眼睛的女人，您将接触到这种类型的内容。

It’s a troubling reality and highlights the need for transparency and collaboration. Involving people from diverse backgrounds can help combat potential biases that get coded into how the algorithm works.

这是一个令人不安的现实，突显了透明度和协作的需求。 让来自不同背景的人参与进来可以帮助消除潜在的偏见，这些偏见被编码为算法的工作方式。

制作相关内容 (Crafting relevant content)

We are still far from the days that AI can create more compelling content than humans, but the gap is shrinking.

距AI可以创造比人类更多的引人入胜的内容的时代还很遥远，但是差距正在缩小。

To get an idea of where AI is today in producing content, I asked Talk to a Transformer, OpenAI’s new machine learning model trained on millions of websites, to tell us “how AI is being used in marketing.”

为了了解AI在当今内容生产中的地位，我问Talk to a Transformer，这是 OpenAI在数百万个网站上接受培训的新机器学习模型，来告诉我们“ AI在营销中的应用方式”。

Here is what it said in response:

这是它的回应：

Artificial intelligence is increasingly being used to sell products. We see this happen at the end of sales teleconferences, with pitches for integrations, recommendation engines, and analytics. It is used to use social media to sell ads. We see artificial intelligence being used to automate processes….

人工智能正越来越多地用于销售产品。 我们认为这是在销售电话会议结束时发生的，涉及到集成，推荐引擎和分析的内容。 它用于使用社交媒体来销售广告。 我们看到人工智能正在被用来使流程自动化……。

Not bad for a machine. Though I don’t know who still uses the term “sales teleconferences”.

对于一台机器来说还不错。 虽然我不知道谁仍然使用“销售电话会议”一词。

Today, most of the uses of AI use in content complement your content marketing initiatives, not replace them. Here are a few examples:

如今，人工智能在内容中的大多数使用都可以补充您的内容营销计划，而不是取代它们。 这里有一些例子：

• Content strategy — Not sure what to write about? AI can be used to help identify which keywords to target and be able to draw out trends/topic ideas, also known as knowledge graphs.内容策略-不确定要写些什么？ AI可用于帮助确定要定位的关键字，并能够得出趋势/主题思想，也称为知识图。
• Content enhancement — Most of the content enhancements are for specific, narrow uses. For example, have your content written but want to improve it or make it more widely digestible? AI platforms make proofreading a breeze. Others can help automate language translation services. One platform helps optimize email subject lines, headlines, and CTAs. Others provide content guidance with quality and relevancy scores.内容增强-大多数内容增强是针对特定的狭窄用途。 例如，您是否编写了内容，但想对其进行改进或使其更易于消化？ 人工智能平台使校对变得轻而易举。 其他人可以帮助自动化语言翻译服务。 一个平台可帮助优化电子邮件主题行，标题和CTA。 其他提供内容指南的质量和相关性得分。
• Content personalization — This can involve modifying ads, content blocks on web pages, or emails to show content that is tailored for the individual内容个性化-这可能涉及修改广告，网页上的内容块或电子邮件以显示针对个人的内容
• Social Media Optimization — AI can help inform what and when to post across various social media channels and when to advertise on mobile devices社交媒体优化— AI可以帮助告知在各种社交媒体渠道上发布什么，何时发布以及何时在移动设备上做广告
• Content communications — This can include automated chatbots, Q&A modules, or even SMS or push notifications delivered at the optimum time内容通信-可以包括自动聊天机器人，问答模块，甚至包括在最佳时间发送的SMS或推送通知

A list of companies that are tackling these challenges can be found on The Marketing AI Institute’s website.

可以在The Marketing AI Institute的网站上找到应对这些挑战的公司列表。

More ambitious uses of AI are working to develop full-written story narratives. According to their customer page, the leader in this space, Automated Insights, currently helps enterprises use natural learning processing (NLP) to automate content for both reporting dashboards and sports recaps.

AI的雄心勃勃的用途正在致力于开发完整的故事叙述。 根据他们的客户页面，该领域的领导者Automated Insights目前可帮助企业使用自然学习处理(NLP)来自动化内容，以用于报告仪表板和运动回顾。

We are still not at the point where computers can write and make sense, as they struggle to understand what is “important” to say beyond finishing my…sentences.

我们还没有达到计算机可以编写和变得有意义的地步，因为它们除了完成我的句子外还难以理解什么是“重要的”。

预测某人何时流失 (Predicting when someone will churn)

Customer churn is like the breakup you didn’t see coming for many companies. It’s painful not only because it typically impacts the bottom line (acquiring new customers is far more expensive than retention), but it also means that the experience was lacking for the customer in some way.

çustomer流失就像你没有看到来了很多公司的解体。 这很痛苦，不仅因为它通常会影响利润(获取新客户比保留客户要昂贵得多)，而且还意味着客户在某种程度上缺乏体验。

Machine learning models can help not only proactively identify which customers are at risk of churning in a period you designate, but also help untangle the possible reasons why. This is a useful tool for B2B, B2C businesses subscription-based or not, assuming there’s enough churn data to develop a model.

机器学习模型不仅可以帮助您主动识别在您指定的时期内哪些客户有遭受搅动的风险，而且还可以帮助您弄清其原因。 假设有足够的客户流失数据来开发模型，那么这对于无论是否基于订阅的B2B，B2C业务都是有用的工具。

Churn data often is a combination of demographic data like location, age, title, firmographic data like industry, revenue, and engagement data call logs, reviews, support requests, product usage statistics, etc.

流失数据通常是人口统计数据(例如位置，年龄，职务，行业数据，行业，收入和参与度数据)的组合，包括通话记录，评论，支持请求，产品使用情况统计信息等。

The result of the machine learning model will be a churn score for each customer, and information on which variables are impacting the model.

机器学习模型的结果将是每个客户的客户流失得分，以及有关哪些变量正在影响模型的信息。

While seeing this information is nice, it won’t make a difference if it’s not used. Intervention practices can include customer support, discounts, and coupons, new product feature announcements, requests for feedback, or therapy. Kidding on the last one, just making sure you were paying attention.

看到这些信息很不错，但是如果不使用它也不会有所作为。 干预措施可以包括客户支持，折扣和优惠券，新产品功能公告，要求反馈或治疗。 开玩笑的最后一个，只是确保您正在关注。

One risk factor with churn prevention may be counter-intuitive. By reaching out proactively to “dormant” customers that might churn, some organizations have seen an uptick in churn as it reminds them of an account they may be paying for but haven’t canceled. Machine learning can help separate which accounts might fall into this bucket as well. In addition to identifying who may churn, similar models can help determine who is at risk of not renewing and who might downgrade their plan.

预防流失的一种风险因素可能是违反直觉的。 通过主动与可能会流失的“Hibernate”客户联系，一些组织看到了流失率的上升，因为它使他们想起了他们可能要付款但尚未取消的帐户。 机器学习还可以帮助区分哪些帐户也可能属于此类别。 除了确定可能会流失的人之外，类似的模型还可以帮助确定谁有不续约的风险以及谁可能降低其计划的等级。

For companies that have implemented churn prevention models, success rates are easy to measure, looking at the churn rate before the intervention and after. Churnly, a churn analytics platform for B2B SaaS companies, promises 92–96% accuracy rates.

对于已经实施了防止流失模型的公司来说，查看干预前后的流失率很容易衡量成功率。 针对B2B SaaS公司的客户流失分析平台Churnly承诺准确率在92％至96 ％之间。

创建更好的报告 (Creating better reporting)

Speaking of measuring, any marketer’s biggest frustrations are proving ROI and gaining full-funnel visibility to optimize spend.

说到衡量，任何营销人员最大的挫折就是证明投资回报率并获得全面的渠道可见性以优化支出。

Today, many attribution models still rely heavily on assigning scores to activity and adding them up to determine relative attribution. Theoretically, advanced analytics can help tie disparate data sources together, and machine learning can help determine what activities to invest more in, and which to invest less in.

如今，许多归因模型仍然严重依赖于为活动分配分数并将其加起来以确定相对归因。 从理论上讲，高级分析可以帮助将不同的数据源结合在一起，而机器学习可以帮助确定哪些活动需要更多的投资，哪些活动需要更少的投资。

Yet, ROI analysis still comes with inherent challenges. Here’s why:

但是，ROI分析仍然面临着固有的挑战。 原因如下：

• It’s challenging to know all of an individual’s brand touchpoints. Some data is anonymous and very difficult to track, such as someone’s first web visit from their home computer, an email forward, etc.了解个人的所有品牌接触点是一项挑战。 有些数据是匿名的，很难跟踪，例如某人从家用计算机首次访问网络，转发电子邮件等。
• We still often assume sales is the last touchpoint and that marketing does not impact the late stages of the sales process.我们仍然经常假定销售是最后的接触点，而营销不会影响销售过程的后期。
• We may not have enough data to figure out which touchpoint was most effective.我们可能没有足够的数据来确定哪个接触点最有效。

Machine learning promises to give visibility into what would happen if marketing didn’t exist and then run what-if scenarios to see what combination of events showed a real impact.

机器学习有望让人们了解如果不存在市场营销将发生的情况，然后运行假设情景，以查看事件的哪些组合显示出真正的影响。

This approach will provide compelling insights on what programs to keep running, and which to throw by the wayside. However, few organizations have the volume and data quality in place across all their brand touchpoints to do this effectively. James Kinley, a data scientist, and Medium writer, wrote an easy to follow guide that can clarify how this works in action.

这种方法将提供令人信服的见解，使哪些程序可以继续运行，哪些程序可以抛弃。 但是，很少有组织在其所有品牌接触点上都具备适当的容量和数据质量来有效地做到这一点。 数据科学家兼媒体撰稿人James Kinley撰写了易于遵循的指南 ，可以阐明其工作原理。

That being said, following similar modeling, machine learning can give visibility into top-performing ads across channels, be it social campaigns, PR, advertising, etc. helping quickly discover which keyword combinations, ad types, and locations are most successful.

话虽这么说，通过类似的建模，机器学习可以让您跨渠道查看效果最好的广告，无论是社交广告系列，公关，广告等，都有助于快速发现哪些关键字组合，广告类型和位置最成功。

One other use not covered in this guide as it spans multiple departments is real-time pricing optimization. This is widely used in the retail and travel sectors. Considering a wide range of behavior, environmental, and timing inputs like competition and season, companies can leverage elastic pricing models to determine initial and promotional pricing in real-time. Some companies have opted for platforms to help facilitate price optimization; others have fleshed out their in-house data science teams.

本指南涵盖多个部门，因此本指南未涵盖的另一项用途是实时定价优化。 这被广泛用于零售和旅游领域。 考虑到各种行为，环境和时间输入(例如比赛和季节)，公司可以利用弹性定价模型实时确定初始定价和促销定价。 一些公司选择了平台来帮助促进价格优化。 其他人则充实了内部数据科学团队。

The question remains, is AI just buzz, or is it the future? AI will be part of the future, but today it’s just not at the level of maturity where it’s beneficial everywhere, for everyone.

问题仍然存在，人工智能只是嗡嗡声，还是未来？ 人工智能将成为未来的一部分，但今天它还没有达到成熟的水平，它对所有人都无处不在。

AI has led us down an unexpected introspective quest, forcing us to confront the question, “what is beneficial?”

人工智能导致我们进行了一次意想不到的内省性探索，迫使我们面对“有什么好处？”这一问题。

As brands, it’s more important than ever to know who you are, what you stand for, and where your boundaries are.

作为品牌，知道自己是谁，代表什么，界限在哪里比以往任何时候都更为重要。

AI is not good nor evil, it’s simply an enabler. While data quality and quantity will remain essential ingredients to any machine learning initiative, the power of AI was never in the data, but how to translate and interpret it into action.

人工智能既不是善良也不是邪恶的，它仅仅是推动者。 尽管数据的质量和数量仍然是任何机器学习计划的基本要素，但AI的力量从来没有出现在数据中，而是如何将其转化和解释为行动。

You can choose whether you want to manipulate people into purchasing more or give people the best possible experience to have them coming back for more. The latter means taking a look at the holistic context of your brand’s relationship with customers before jumping headfirst into an AI-first strategy.

您可以选择是要操纵人们购买更多东西，还是要给人们最好的体验，让他们回来购买更多东西。 后者意味着先了解品牌与客户关系的整体背景，然后再优先采用AI优先策略。

As a marketer myself, I am incredibly optimistic that the future of AI will not only help companies get a leg up on the competition but also deliver better experiences for everyone.

作为我自己的营销人员，我非常乐观地认为，人工智能的未来不仅将帮助公司在竞争中立于不败之地，而且还将为每个人提供更好的体验。

Perhaps the key to long-term success in AI and marketing can be captured in one sentence, “Do your marketing like humans like to be treated.” The most successful organizations will be the ones that foster collaboration between both data and marketing leaders to determine the best path forward.

人工智能和营销长期成功的关键也许可以用一句话来概括：“您的营销是否像人类一样喜欢被对待。” 最成功的组织将是那些促进数据和市场领导者之间的协作以确定最佳前进道路的组织。