如何利用人工智能来增强人类的智力

从人工智能在短期内取代人类的工作, to the singularity-induced creation of super-intelligent killer machines in the long-term, seldom a day seems to pass without AI drawing public criticism. 然而,, even as our alarmist media preys on humans’ most basic fears, AI is silently improving our lives one breakthrough at a time. 

German philosopher Immanuel Kant declared that there exists a world which humans cannot conceive of, 这是我们无法理解的. Now the computational power of AI is providing tantalising glimpses of that world. 一个完美的例子是A 最近的MIT项目 in which AI discovered a new antibiotic that is now successfully killing antibiotic-resistant bacteria. 

The AI succeeded by screening more than 100 million chemical compounds in a matter of days and identifying patterns that were indistinguishable to humans. 确定了这些模式存在的, 人类研究人员现在正着手解释它们, and in so doing are expanding the human race’s understanding of molecular cell biology.  

以这种方式, 人工智能和人类是共生的, with the AI unlocking doors through which human understanding can step. It is this combination of AI computing power and human rational thinking and interpretation that provides the opportunity for an AI-powered society that is massively enhanced in many fields - and advertising is no exception. 

Advertising adopted AI technology with the aim of optimising content. Several systems have been created that can ingest content and data, and determine the most effective cut of an ad; some systems even automatically cut the ad into its “optimal” version. 然而, many of these systems have left advertisers feeling short-changed; complaints that the outputs lack nuance and are one-dimensional abound. 

Herein lies the challenge of AI – an AI system is only as good as the data on which it’s based, so a system trained on shallow social data or basic media metrics will only ever produce shallow, 基本结果. However, when the AI is trained on deeper data, the outputs can be magical. 

The moral of the story is to check what data your AI has been trained on. Do not accept vendors’ claims at face value and always be willing to dig deeper. 特别要注意的是: 

1. 被学业或证书弄得眼花缭乱. Many AI-based techniques are backed by academic research or created by companies with academic founders. While this can certainly add knowhow and credibility to a company, AI is advancing so quickly that core characteristics of a good AI practitioner include humility to know that they don’t have all the answers and open-mindedness as regards new techniques and data sources. Any AI practitioner without these characteristics should be treated with extreme caution.
2. 声称能识别人类普遍真理的算法. If an AI vendor claims that all humans everywhere respond to a certain stimulus in the same way, it is very likely because it suits them commercially to make this assertion. 通过宣称发现了一个普遍的真理, vendors actually reveal that their model has been trained on a small and/or shallow dataset, and that they are inferring that larger populations/different cultures will respond in the same way. 在现实世界中，他们很可能不会! 
3. 基本输出. 当人工智能系统只能产生基本输出时, 这意味着训练它的数据是基本的. Basic data inputs lead not only to oversimplified outputs, but inaccurate ones. 再一次，买家要小心了!  

In my experience the best AI systems are run by teams who continually add new 培训 data and analysis techniques to their repertoire to increase the accuracy of their predictions. 

应该使用哪些训练数据? 越深越细越好. Some of the most interesting data to be collected in advertising in the last five years is attention and emotions data. 通过摄取这些数据, AI can reveal to advertisers why their content works or doesn’t, 从简单的相关性到因果关系. Only when this holistic understanding of content performance is achieved can advertisers learn from the past and optimise strategy for the future. 

总之, AI is creating huge breakthroughs in just about every field of human endeavour, 而不是通过取代人类的智力, 而是通过增广. Advertisers should embrace AI technology but must also beware – not all AI is created equal. 判断一个人工智能系统是否适合你, really get to understand the data on which the AI is built. 不要被学术研究弄得眼花缭乱, be wary of algorithms that appear to unearth universal truths and walk away when presented with basic outputs.