When we hear the term ‘artificial intelligence’ our minds automatically turn to a world ruled by robots, but AI is a growing resource that has many practical implementations, many that span beyond the realm of digital marketing.
As our awareness of artificial intelligence grows, the hype around AI will diminish as it will become an expectation in our everyday lives, much like electricity and the internet.
It is useful to have an understanding of how AI works and its various applications in everyday life, as well as how it might shape the future.
In this article, we provide an explanation of several types of artificial intelligence along with working examples of how they are being used.
Noun: "the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages."
Explanation: Machine learning is an application of artificial intelligence that equips systems to automatically learn and improve from experience without being explicitly programmed.
The process of machine learning uses observations or data, such as examples, direct experience, or instruction, in order to recognise patterns in data that we provide to make better-informed decisions in the future. Machine learning aims to let computers learn automatically “for themselves”, eliminating the need for human intervention or assistance, so systems can adjust their actions accordingly.
By giving a machine lots of examples to analyse and learn from, it will develop generalisations that can be applied to new situations. Statistics mean machines are able to use algorithms to identify patterns and make decisions.
Example - Google Lens:
Google Lens enables users to search what they see using their smartphones. For example, you could search for a particular kind of flower. Google will return a search result telling you the type of flower and some information about it, and also suggest local florists that you may like to order this flower from.
Ever liked your friends outfit? With Google Lens, you can take a photo and Google will return search results of similar items so you can recreate the look.
While there are not currently any published guidelines for optimising products or images for image search, Google is committed to providing the end user with the best possible experience, so it is likely that as this form of search grows, there will be ways to improve your search visibility to compete with other sites.
Example - Google Ads:
Google Ads is able to make suggestions for optimising ad campaigns based on previous campaigns. It analyses data including ad headlines, descriptions, extensions and other relevant information such as landing pages to make recommendations to increase the number of conversions and return on investment (ROI).
Google Ads Machine Learning Solutions offer the following four ways to simplify and grow your campaigns.
1. Targeting: so you can acquire the best customers via dynamic campaigns.
2. Creatives: improving creatives by using the right message for every moment, for every user.
3. Bidding: extending the lifetime value of your customers by reaching them more efficiently by adjusting bids in real-time with automated bidding.
4. Attribution: optimise your campaigns for growth by taking a holistic view and bid beyond the last click.
The advances that Google is making with machine learning does not mean marketers can simply “set it and forget it”. Artificial intelligence is enabling humans to gain better insight by analysing quantitative and qualitative data and ultimately make smarter decisions. AI in Google Ads frees up time for account managers and marketers to focus on higher level decisions such as campaign strategies.
Explanation: Deep learning is an advanced sub-category of machine learning. It is the processing of large amounts of data, including abstract and scattered data, to discover complex patterns and correlations that can be used to understand a consumer’s interaction that leads to better individual targeted campaigns and ROI.
Example - Amazon:
Have you ever purchased a book from Amazon to then be faced with “People who bought this book also bought this book” or “Recommendations for you”?
Amazon uses a sophisticated algorithm that uses deep learning artificial intelligence to analyse huge sets of data to make educated product and deal suggestions based on your activity. The online retail giant has enormous volumes of data that not only monitors your buying habits but the buying habits of others who have bought products similar to you. Amazon’s complex algorithm uses this information to make educated product predictions that are specific to you.
This is a great example of consumer insight being explicitly analysed to personalise the user’s experience, increase customer engagement, drive sales, and build brand loyalty.
Explanation: A neural network is a combination of algorithms for different tasks that feed into one another to process data, recognise patterns and trends, and learn to develop strategies. They’re modelled to mimic the human brain and how it processes information.
Neural networks learn from experience, much like we do as humans.
Like the human brain, the more experiences (data) a neural network processes, the better they become. The benefit of neural networks are that because they learn “organically”, they can self-repair and correct what they have learnt.
Neural networks are developing dynamic tools for marketers, allowing us to process large sets of data that give greater insights to help better predict consumer behaviour, creating and understanding more sophisticated buyer segments, marketing automation, content creation and sales forecasting.
Predictive analytics allow marketers to predict the outcome of a campaign by recognising the trends from previous campaigns. While neural networks have been around for some time, there is a greater demand to process Big Data so systems are becoming much more dynamic and intelligent as a result.
Example - LinkedIn:
In 2014, LinkedIn acquired job search start-up company Bright.com to offer better job-candidate matches for both employers and job seekers.
Bright uses machine learning algorithms that use data from user profiles to improve the user experience. These sophisticated algorithms consider user’s historical hiring patterns, account location, previous work experience, and synonyms in job descriptions. Once Bright has completed its analysis, the system assigns a Bright Score that indicates the quality of the match between the job role and the candidate.
Explanation: Natural language processing is a machine's ability to understand and process the meaning and context of a human’s voice, be it by text or speech.
The human brain continues to build on its experiences to process natural language. Whenever you read an article or hold a conversation, you have the ability to process the words, understand their meaning, feel emotions about the subject, and are able to visualise how that thing would look in real life.
Text and speech recognition presents a long list of problems when it comes to machines learning how to process natural language. Systems have to process copious amounts of data to develop an understanding of the quirks and subtleties in the human language.
Text characteristics include:
Speech characteristics include:
Natural Language Processing is extremely complex as not only does the machine need to understand the above characteristics of language, it also has to learn how to decipher overlapping speech and accurately understand speech in noisy environments.
NLP often plays a role in biometric authentication where AI recognises unique characteristics of a user’s voice. NLP is commonly used for automated phone systems, fraud detection and Chatbots.
Example - Home Assistants:
If you don’t have a home assistant already, you’ve probably got a friend or family member who does. eMarketer predicts that the use of smart speakers in the UK will grow by almost a third in 2019, after doubling in 2018.
Home assistants use natural language processing to perform instructions or actions from the user. This may be ordering a takeaway, asking what the weather is like or playing a song.
Example - Chatbots:
Advanced Chatbots are powered by artificial intelligence to help the system:
Without NLP, Chatbots would not be able to differentiate between ‘Hello’ as the start of a conversation, and ‘Goodbye’ as the end.
Chatbots provide consumers with an additional touchpoint for transactions that are readily available at every step of their journey.
Explanation: As marketers, we’re already used to engaging with data analytics to identify trends and patterns, find relationships between variables, and scope out the areas of opportunity. But, it does not predict the impact of a change in a variable. Humans can only make relatively broad assumptions about future patterns and outcomes when analysing data. Data analysis summarises trends and patterns from past events, so it is up to an individual to decide what this infers about the future.
The introduction of artificial intelligence will change the way we use data analytics thanks to its ability to process Big Data. AI machine learning means the system is able to make assumptions, test, and learn autonomously to build models that deliver predictive insights that marketers can use to build more effective campaigns.
The use of artificial intelligence in the digital landscape is changing the way we market. AI is giving marketer’s the tools we need to process Big Data and gain better-educated insights. In turn, we’ll be able to build better campaigns, deliver more personalised user experiences and analyse improved insights to help make future decisions. Artificial intelligence is able to process data at rapid rates that cannot be rivalled by humans. It gives us the ability to focus our efforts at higher levels in terms of developing strategies, building targeted campaigns and effectively managing data so that it can be applied to future strategies and decision making.