AI: one technology to rule them all
AI assistants are coming. It’s time to rethink your company’s value proposition.
Product recommendation personal assistants based on artificial intelligence (AI) will help people get information, negotiate and make transactions. This will have significant effects on marketing, sales and operations in many industries.
Widespread use of these tools would cause a loss of relevance to companies and brands by reducing direct contact with their customers. The resulting increase in competition would benefit consumers and make a huge difference in selling a majority of consumer goods and services.
Scalable AI software able to serve millions would exploit that capability to improve their feedback to individual users, attracting new ones in the process. This network effect means the AI service would become more attractive as additional users join.
Unless a free AI scenario prevents it, the above would lead to huge increases in valuation for market leaders and, conversely, to big losses for companies in the long tail. If the AI leader is a startup the term unicorn wouldn’t do it justice and we might be having to come up with a new word.
This is kind of a long article. Here is a table of contents so you can navigate it back and forth. In case you don’t have much time I suggest you read from point 5 onwards and come back later.
1. Where we are today
2. Purchasing process: present and future
3. Impact on corporate strategy
A future where the tangible reigns / Informed decisions / Signing-up… and canceling / Cross-selling and ecommerce / Operational challenges and opportunities / Resistance to change
4. Every industry
Hypermarket and other big retail / Shopping malls / Advertising / The end of app stores? / Influencers and trendsetters / There’s more
5. Big Tech and the war on surplus
Companies try to set themselves apart from the competition, in an effort to increase their margins and market share. To meet this goal businesses make use of multiple resources: technological, operational, financial… One key discipline in management is marketing: a set of activities and processes necessary for a company to meet the needs of its customers and induce them to modify their behavior for the firm’s benefit. Marketing includes, among others, advertising, distribution, selling techniques, pricing, product design or brand development. In short, it’s a big deal.
As potential customers we receive a continuous bombardment of commercial messages, from all types of media and in different formats, trying to get us to buy a particular product. In fact, in today’s society the acquisition of goods and services is a routine task. Being true that, for many, shopping has become kind of a hobby, it is also an activity that demands a fair amount of time and effort, and carries substantial costs.
The aforementioned model has remained almost unaltered for decades, but could undergo major changes with the upcoming introduction of AI-based software. Throughout this article I may indistinctly refer to this kind of software as AI product recommendation, software, system, bot, shopbot, personal assistant, AI, advisor, personal shopper, chatbot, etc. You may think of them as a natural evolution of software assistants (Siri, Cortana, Alexa…).
Leaving impulse purchases aside or those cases for which there are no product alternatives, we can distinguish several stages in the buying process:
- Setting up of initial consumer preferences (features the product should meet) together with the corresponding budget constraint.
- Collection of information: on the product, substitutes, offers, where/how to buy, financing, etc.
- Search for feedback from other consumers.
- Transaction, including negotiation if applicable.
- Consumption of the good or service. It may involve making use of an after-sales service or guarantee.
- Finally, and taking into account all of the above, buyers will make an internal assessment of their experience, which may lead them (or not) to buy again in the future and will affect their feedback to others.
Each of the above phases takes time and costs that could be significantly reduced by the introduction of AI-based product recommendation systems. How would this kind of software work?:
- Once we feed the software with our preferences the system would use them to focus its search for offers. Voice-user interfaces and natural language recognition software would make this easier.
- A personal assistant would also use preferences not directly disclosed by us by analyzing our online searches, purchasing habits, etc.
- Collecting information from numerous sources shopbots would cover much more ground than us in a fraction of the time: news, online merchants, catalogs and listings, forums, reviews of buyers in online stores, etc. For example, an exhaustive analysis of reviews could filter out bogus ones, which only seek to promote a product or harm a vendor.
- The system might even evaluate the publicity we get when we watch TV or browse our computers and phones. Yes, even the fine print running at the bottom of the screen.
- AI advisers would be immune to human cognitive biases, cleverly exploited by brands and widely used in pricing and product communication. This would save us money and allow us to focus on the actual features of the product.
- In the face of a huge (and growing) offering our AI would present us a limited selection of options, tailored to meet our budget and preferences, so that we can make an informed final decision.
- Over time we could be leaving some purchasing decisions entirely in the hands of these systems.
In order to do all this, shopbots would take advantage of the fact that increasingly more product information is available online.
Recommendation software would also make easier evaluating and comparing products. If some time after purchasing some product the AI asks me to evaluate it my rating would be stored and could serve as a basis for comparison for future purchases.
This ability to store feedback could be used to include additional information: on obsolescence (when a product stopped working/broke), quality of after-sales service, etc.
Apart from finding and comparing offers and protecting us from our own biases, the AI would also provide negotiation tips for certain types of purchases.
On top of the above the aggregation of preferences for many users would make possible arranging joint purchases in order to get quantity discounts.
In summary, the potential benefits to the consumer of these virtual personal shoppers would be multifold: time savings (after discounting time dedicated to interact with the system), cost savings, avoidance of cognitive biases, better decision making thanks to reduced complexity (comparability of offers, selection/prior screening), discovery of new products, etc.
AI systems could also be used to manage household stocks by working with inventory levels, obsolescence and expiration dates, reorder points (including automatic purchase or subject to user approval), etc.
The potential for improvement over the current situation is significant. As an example, a recent study from OCU, a Spanish consumer organization, revealed that Spanish households, on average, could save up to €933 every year depending on where they made their daily shopping (as a reference that amount represents 1/5 of the annual expense).
As product advisors improve their ability to communicate with the user, become invisible and prove their usefulness, adoption rate should pick up speed until they become ubiquitous.
The scenario described so far would optimize the value or utility consumers get from their shopping basket by…
- Reducing their expenses and as a result increasing their purchasing power (budget constraint).
- Modifying users’ preferences with the suggestion of new products more to their liking and preventing people from buying again products they didn’t like in the past.
As a result consumers would get a higher level of satisfaction for a given budget constraint and a budget surplus, having the option to use it to increase their consumption and/or their savings. This surplus concept is relevant and we’ll get back to it in section 5.
The possibilities opened by AI are immense. As Enrique Dans writes in Right Now, Artificial Intelligence Is The Only Thing That Matters: Look Around You : “[…] machine learning and artificial intelligence are the keys to just about every aspect of life in the very near future: every sector, every business.”
As we will see, more specifically the use of AI to get information, recommend products, negotiate and carry out transactions may have formidable consequences in marketing, sales and operations. Business models will be affected and strategy will have to change accordingly.
AI product recommendation promises to shift the conversation about products, taking us to a scenario where tangible arguments would carry more weigth than intangible ones (the ones mainly used by advertisers). This would be good news for the consumer and would make a huge difference when it comes to designing and marketing products and services in many industries.
Companies seek recognition for their brands, an identification of the audience with their products that allows them to differentiate themselves from the competition. This in turn helps companies charge higher margins while promoting customer loyalty. The upcoming emergence of AIs as intermediaries between brands and consumers poses a serious challenge for companies.
When the focus is on the tangible qualities of a product, when you must convince some software that your product is superior to the competition, then it’s time to rethink your value proposition, starting with the design of your product. As brands lose relevance, constant innovation in products and processes becomes critical.
The increased price elasticity of demand resulting from the introduction of AI product recommendation (customers become more price sensitive) will make life harder for companies and might lead to a commoditization of many products and services (conversion into undifferentiated product). Companies will make great efforts to retain their customers and after-sales service may gain importance as a result. However, the implications of AI systems do not end here.
When making a purchase decision it is usual for consumers to miss some information, and makes them underestimate costs or overrate benefits. In fact, product information from sellers does not normally include all related costs. There are fine print and hidden costs (indirect, maintenance) that are not considered by final customers due to a lack of knowledge or incentives to analyze them (companies do have proper incentives and knowledge).
Product recommendation systems would take into account most of said costs and, by aggregating information from many buyers, would determine the actual cost of a product to be used later as an input in the buying process.
In most cases information asymmetry between buyers and sellers benefits the latter, allowing companies to charge higher margins. The application of AI systems would mitigate this asymmetry, resulting in further price reductions.
In some cases, however, we may find the reverse situation, where buyers have more information than sellers. When this occurs, and to prevent adverse selection, companies try to push potential customers to disclose information about their situation and preferences in order to present them with an offer that maximizes company’s profit. Quite probably these personal assistants will make companies miss some information customers used to reveal.
Signing-up and cancelling
Services (financial, insurance, utilities, telecommunications, travel, real estate, etc.) are another field of application of AI. Such application would range from a simple search for offers and estimation of the necessary capacity (the system could make this calculation based on information about our consumptions, income, expenses…) to the signing up and, eventually, termination of the service.
In order to sign-up for some service, users must agree to long terms and conditions written in legal jargon that, if online services serve as an example, almost nobody reads. In order for AI to read, understand and even accept such terms on our behalf, it would be necessary to simplify them, to convert them to a standard model. This standardization could be carried out by service companies themselves (motu proprio or forced by new legislation) or by third parties (AI suppliers, consumer associations, etc.).
AI’s ability to easily terminate services is another factor to consider, particularly since cancelling is always harder than signing up. This lowering of barriers to exit stresses the importance of keeping existing customers. An AI system could also prevent automatic renewals and alert users of future changes in conditions, thus increasing the bargaining power of customers. The same would apply to the exercise of the right of withdrawal.
Cross-selling and ecommerce
Shopbots pose a considerable threat to cross-selling, a key element for some business models. In the future, customers may indeed let their recommendation software explore complementary products. And that would be their decision, not the default option as it is now.
Cross-selling allows companies to sell high-margin products and services by taking advantage of an existing commercial relationship, using other products as a hook, in complex sale processes or when selling high-priced products (e.g. travel insurance at the time of booking a flight). For the case of services, cross-selling is particularly important because it helps retain customers by increasing barriers to exit.
Cross-selling is crucial in e-commerce, given the information available on user preferences, inferred from their browsing habits and purchase history. Later merchants use that information to implement cross-selling strategies and design customized offers.
But, what if an AI system uses anonymous browsing and hides customers’ preferences? What if the software collects information from some database instead of visiting the merchant website? AIs could delete cookies and browse from corporate servers, masking identities and repeated visits and sharing product information from websites with millions of users. There may be ways to mitigate this problem, but this is no laughing matter. In fact, shopping bots might render useless many of today’s most popular online marketing strategies.
Operational Challenges and Opportunities
The use of AI product recommendation software by consumers would very much impact the operations of manufacturers and sellers:
- Price changes made by a single seller can induce unanticipated changes in demand. As AI software modifies their recommendations accordingly, some companies might be unable to meet demand for the product or risk accumulating unsold stocks.
- The same could be said for changes in product features. For example, a brand’s decision to stop using palm oil in their cookies migh lead to a spike in demand, as the new configuration matches the preferences of a larger set of consumers. Increased demand could be offset by those no longer interested after the change.
- Group buying: AI systems might pool buyers in order to get volume discounts. Manufacturers might have to deal with this type of order more often in the future.
- Product recommendations based on parameters and preferences would drive consumers to try out new products and vendors. An opportunity for new entrants and niche manufacturers, as this creates a more level playing field.
- Bandwagon effect, fashion trends. Having information from millions of users, the software can detect buying trends for a particular product, which may lead to bandwagon or fashion phenomena.
Broadly speaking, manufacturing and logistical flexibility will be key to compete and seize opportunities in a future with wide-spread AI technologies. The increased uncertainty and variability in the new scenario can be reduced, among others, following these lines of work:
- Creating and communicating a more tangible value proposition.
- Treating data as a fundamental asset of the company (data capital concept).
- Usage of recommendation tools to track competition and fine-tune marketing and sales strategies (think as a customer).
On the other hand, in the new environment there would be possibilities for both existing companies and new entrants. For example:
- Development of standards for information on products.
- Creation of databases that AIs can access to collect information about products/services.
- Search services for offers: subscription-based, one-time payment…
- Design of algorithms for product evaluation, generic or category-specific. They could be sold as packages for our personal AI.
- Certification agencies that assess websites’ safety and compliance with standards and good practices for working with AI (“AI friendly” tags). This might be important to give consumers confidence and to avoid fraud.
We must not forget here the opportunities for AI software providers of having such a vast amount of consumer data. Brands and merchants could make good use of these recommendation engines themselves or buy access to aggregate data to optimize their offer and strategies. Something similar can be said of market research firms if they are to stay in business.
Resistance to change
The outlined scenario does not sound particularly encouraging for many companies (and their managers), confortable within a model and environment that they know too well. Thus, resistance to change is to be expected. Here are some possible examples:
- Proprietary data: Companies may argue that the information on purchase terms and product features is proprietary and cannot be extracted, aggregated or treated without their consent. The intent is clear: to prevent AIs from collecting it and avoid the emergence of databases and other business models built around this info.
- Protection of personal data, privacy. The reasoning goes like this: AI systems would have access to all our data, some of it certainly sensitive (financial, browsing, location), which represents a serious risk (in case of misuse, theft, identity theft, etc.). Apart from telcos and other companies this argument could be shared by many consumer organizations.
- Requirement of human intervention to prevent software from signing up on our behalf. This is a similar argument to that claiming that law requires vehicles to carry a human driver, employed by the auto industry against the threat of autonomous vehicles.
- Pressure to limit joint purchases, on the grounds that they favor large companies and can lead to dumping.
- Trade unions and professional groups: lobbying actions to mitigate and/or delay the possible impact of product recommendation software on employment (sales forces mainly).
Furthermore, the introduction of AI software will create incentives to beat the system. Programmed purchases that run automatically when certain criteria are met could yield juicy profits for those who detect a weak point, which will lead many to try to ‘fool’ the AI. Not unlike those bogus reviews, ambiguous descriptions and inflated prices offered by some sellers at sites like Amazon or eBay.
Certain level of fraud seems inevitable, but standards and certifications could help in this regard. Also, the exchange of information between AIs will contribute to build an online reputation for merchants that would progressively help markets to get rid of bad sellers.
One way or another all industries will be affected by AI, which will have a particularly significant impact on brands. It is worth considering how it might affect some sectors.
Hypermarkets and other big retail
The supermarket and hypermarket formats are losing their appeal as a result of the emergence of ecommerce, fueled by the improvement of logistical processes. On top of that we have the upcoming revolution of autonomous transportation or the possibility of serving remote geographic markets using drones. AI would only exacerbate that trend:
- Product recommendation systems would take care of our everyday shopping. To do so the software would use our feedback and keep an eye on our stock of groceries and other essentials.
- A more than likely drop in revenue from cross-selling would be expected. Supermarkets aisles are designed to take advantage of our biases and encourage us to buy products we do not want/need. AI doesn’t have any of those.
As if this was not enough Amazon is entering the everyday shopping arena, where its strenghts and capabilities could help it dominate the sector in the new scene drawn by AI:
- Top reputation in online sales.
- State-of-the-art technological and logistics platform.
- An increased investment in its own private label products. That would fit with the scenario pointed out before (tangible vs intangible). Bad news for brands.
- A very important investment in AI, with the launch of products such as the Echo family that are being replicated by some competitors.
These are difficult times for brick and mortar businesses. Pressured by the rise of ecommerce, some of their stores are becoming uninteded showrooms, which consumers use to check products physically before buying online from their computers or smartphones (and too often from a different vendor). This is happening against the will of merchants, but we are beginning to see it as a conscious strategy for some of them (Zara — Inditex).
The future doesn’t look pretty for shopping centers, impacted by several factors, AI being one of them:
- Traditional anchors are losing relevance: big department stores are in decline, while cinemas have to compete with home cinema systems and VoD platforms. Others like supermarkets will suffer with the emergence of automatic purchases brought by AI product recommendation software.
- AI software will reinforce the trend towards the showrooming of offline stores. And by adding an extra of rationality in the purchasing process a reduction in impulse buying may be anticipated.
- Augmented and virtual reality technologies will bring advertising, product information and sales off the shopping center to wherever we are.
- Increased use of payment methods other than cash.
On the other hand there are factors that could favor shopping centers, such as autonomous transportation, which may make easier and less costly visiting the mall (however, at the same time this will improve the logistic capabilities of ecommerce). Also, future changes in the labor market could increase the amount of free time available and get people to visit malls more often.
We’ll see how all these changes affect the value of commercial real estate. Said value is based on the rent of the leasable area, resulting from the traffic of the center and the application of spend rates. If traffic falls and the spend ratio gets worse that would cause a significant depreciation of the asset. To survive shopping centers will have to reinvent themselves, changing their offering to become places of leisure and socialization, selling experiences and holding more live events.
The advertising business may be one of the most affected by the introduction of AI personal recommendation engines. Traditional formats and channels have begun to give way to new ones, although all of them are still trying to reach consumers’ heads. New developments that undermine the industry are the gradual disappearance of the television grid, based on advertising segments, which is being replaced by ad-free VOD (Netflix, HBO…) or the use of ad blockers in internet browsing.
In a future in which the AI comes between advertisers and audience, those on advertising will need to devise new formulas to skip this new middleman. We are likely to see a boom in product placement in films, television, video games and literature, as well as growth in sponsorship and non-advertising communication (looking to promote demand for certain product configurations and specifications). Metaphorically speaking (or maybe not) our personal assistant software might very well end up watching ads on our behalf, trying to separate the wheat from the chaff for us.
As we saw before, access to pooled data collected by recommendation software will be very useful to market research companies, performing campaign analysis, pricing strategy, etc.
Upcoming technologies such as augmented and virtual reality will open the door to new opportunities to reach consumers, but all in all the advertising industry is in for a bumpy ride. Google, the undisputed leader of online ads and search, might be in real trouble as we will discuss later.
The end of app stores?
Last year, following a post from Marco Arment, I wrote a post in Spanish outlining the risks AI could bring to the current application model. The article’s point was basically that AI would make apps become invisible to users, leading to an increased standardization of devices. That in turn could harm the margins and relevance of manufacturers like Apple.
The current application-based system has significant limitations and disadvantages:
- Different apps needed for different uses.
- Users need to interact constantly with the applications to get what they want.
- Whenever a new use is proposed the user must look for the appropriate app. Add to that a constant search for “something that works best”.
- Time and hardware consuming technical issues: installations and uninstallations, updates, memory, etc.
AI software will supposedly try to anticipate our needs, reducing our investment of time and effort as a result. This can lead to apps becoming invisible to users and affect the way those are monetized. We could go, for example, from paying for an app to a subscription model in which we would pay a fixed amount for certain service level.
In this new scenario our devices’ AI software (developed by the device’s manufacturer or a third party) would deal with a multitude of apps to get the information or service we need/want (ideally even before we know it ourselves).
The use of AI software in our smartphones could be a serious threat to the current application paradigm. If the AI is devoted to finding the best solution for the user, this has very important implications for the application business:
- Being invisible to the customer the competition among apps gets much tougher. Although theoretically this would favor meritocracy, it would make necessary to develop a system for measuring and ranking based on results obtained and users’ feedback: what is the right app to answer this request according to the user’s preferences and constraints?
- Many applications make money with advertising. This business model is doomed if apps are running in the background. Those would be bad news for developers, who must rethink their monetizing strategies.
- There are more risks for app developers: AI owners might charge them a fee for their apps to be included in the AI scope; AI owners might favor their own apps.
- In order to retain users for their AI platforms and/or devices, companies may try to close exclusive agreements for killer apps (similarly to TV rights deals).
Moreover, if applications are installed, run and updated remotely, they would provide the best results to users regardless of the user’s devices, which would become less important than they are now. It is true, however, that this would require a very fast and reliable connection and at a reasonable price (will 5G be the solution?). It neither takes into account hardware requirements for other emerging technologies such as virtual and augmented reality.
A company such as Apple cannot afford the risk of standardizing devices and will strive to maintain the relevance, share and margins of their product lineup. Other manufacturers such as Samsung depend on others’ operating systems and might very well end up providing commodity devices for AI software to run.
Influencers and trendsetters
It is common practice in many industries to use experts and personalities to promote sales. Although many are known for their professional activity or their status as celebrities, lately more influencers build their reputations in social networks. This newly acquired fame places them on the radar of brands, which end up hiring them to recommend their products. These ‘salespeople’ can bring significant sales, although their effect can sometimes be difficult to measure.
AI software would give prescribers and trendsetters a chance to get the most out of their ability to influence others. A recommendation system could allow users to follow certain influencers and consume those products they buy or recommend. Furthermore, being able to make purchases through the system could provide very valuable metrics to sellers. On the other hand these metrics would also help influencers to put a price on their advice when negotiating with brands, and even to make decisions about whether or not directly entering a business or creating their own brand. While this possibility moves away from the more rational approach promised by recommendation engines, it will probably come true very soon. In fact, it could help a company such as Twitter come up with a profitable business model.
As discussed before, many types of services would be affected by the introduction of AI technologies. Banks, telecoms or energy companies come to mind, but there’s more: real estate, travel, insurance. Too often these industries rely on big information asymmetries to earn money from retail customers. Recommendation software would help users select the product that suits them best and to avoid unnecessary fees and additional services they don’t need.
Lately the prevalent paradigm of buying/having is being replaced by that of renting/using. Good old products are being turned into services, which are a good match for shopping bots. Take the case of auto making, which faces a potential threat to its existing business model driven by autonomous transportation. Self-driving vehicles would provoke a shift from the current model, selling cars, to another based on providing transportation services. That would bring the industry within the scope of AI systems with all the implications pointed throughout the article.
As stated before the potential of artificial intelligence is huge. It may lead to the emergence of new business giants and determine the fate of some of the most relevant companies to date. While for some, AI threatens their business, for others it is a great opportunity to make [a ton of] money. And it may be both for most.
David Paul Morris | Bloomberg | Getty Images We will "see more technological advances over the next ten years than we…www.cnbc.com
Product recommendation systems in particular bring with them an enormous business opportunity that many companies will try to seize. What is it?
Following a simple model from ‘Economics of Strategy’ (David Besanko et al.) let us begin by defining the economic value created by a product (EV) as the difference between the benefit received by the consumer (B), i.e. the maximum price the consumer would be willing to pay for the product, and the cost of said product (C):
EV = B - C
If we add both P and -P (where p = price of the product) and rearrange the expression we arrive at:
EV = (B - P) + (P - C)
That is, the economic value created by a product is the sum of the consumer surplus (B - P) and that of the producer (P - C).
In a scenario prior to AI, transaction costs and disinformation and noise brought by advertising reduce the consumer surplus to the benefit of the producer, by approaching P to B (you will notice the surplus for a company is maximum when B = P ).
However, as AI product recommendation systems lead to an increase in price competition, companies are forced to lower their costs in order to maintain their margins. In this new scenario firms are not able to capture a big chunk of consumer surplus anymore and some may even face serious difficulties to stay afloat.
Also, there’s a new player in the market, AI service providers, who might choose different business models when entering:
- Charging companies a fee for every transaction made. Businesses would find very difficult to pass on this cost to their customers while facing stronger price competition themselves.
- Charging marketplaces. These could in turn pass on this cost to manufacturers.
- Charging software users: subscription fees, transaction fees, etc.
- Advertising model: this however contradicts the AI’s promise of increased rationality.
- AI is provided free of charge and revenue comes from cross-selling, sale of aggregate data, etc.
- Mixed models, others.
Entrants will try to position themselves between consumers and brands in order to capture part of both surpluses. Widespread use of this technology can be very lucrative. For instance, who owns the customer in this scenario?
As a result of the above many will be interested in leading the AI market. Provided no unknown startup prevents it, big tech companies will fight to earn that spot. Businesses from other industries, all of them facing some disruption themselves, will try not to lose this train: telcos, payment service providers, banks…
Some names we know too well: Alphabet (Google), Amazon, Apple, Facebook, Microsoft. None wants to miss this opportunity. These companies are fully committed to an AI future and have launched or are working on AI-related projects. In fact, all of them involve personal assistant software that might be equipped with the abilities laid out throughout this article. Some are also working on new devices or enhanced versions of current hardware.
- ALPHABET (GOGLE): its revenues come mostly from advertising, a source of income seriously threatened by AI (Daniel Colin James made a compelling argument on the problems the company might face soon). It’s not just the fact that people are losing interest on ads. If users no longer need to search themselves and start leaving that to software, the value of the very core of Google, its search technology, might collapse.
- Alphabet is heavily invested in AI, but also in other technologies such as virtual and augmented reality that could generate new advertising opportunities in the future and help the company diversify its income (some possibilities here). Another opportunity might arise from collecting and preparing information for bots to mine. If Google manages to become the preferred search provider for bots, the de facto standard, it could retain part of its relevance and value. Problem here would be how to make money in that situation.
- AMAZON: the company enjoys a privileged position on account of its technological, logistical and commercial capabilities, which should grant it a prominent role in the future. In fact, Amazon could accelerate adoption of software advisers by imposing standards on product information to its thousands of providers, forcing competitors to follow suit. Also, since its own sites are an ideal training ground for AIs, it makes sense to think that, at any time, there are AIs (Amazon’s and others’) running through their pages collecting information, making purchases and fine-tuning their algorithms. The Alexa/Echo product line, that was welcomed as a minor product at the time of its lauch, is gaining traction and serving as inspiration for competitors.
- APPLE: The sale of its devices is critical for this company and, as previously mentioned for the case of the application model, that revenue stream could be affected by the irruption of AI-based technologies. Device commoditization is a too big a threat and Apple is said to be working on a dedicated chip to power AI on devices. Also, at its next keynote, the company is set to unveil a Siri speaker that could be very well a competitor for Amazon’s Alexa/Echo lineup.
- FACEBOOK: the company faces a situation similar to that of Google. Its dependency on advertising is even higher, so Zuckerberg and his team will have to find different sources of revenue to make up for a likely drop on ads in the medium term. Its huge user base, estimated at 1.7 billion at the time of this writing, together with that of Whatsapp, generate a massive amount of valuable data. This is a tremendous asset to be [further] exploited by the company, but it does not guarantee its survival.
- MICROSOFT: Its revenue mainly comes from the sale of services and licenses, and less than a year ago the company acquired LinkedIn to stand its foot on social networks and to tighten its grip on the corporate market. A future with applications running in the background could hurt the Redmond giant, although its reliance on corporate customers could cut Microsoft some slack.
There is an additional possibility: a future with free AI. Mycroft, a startup is offering an AI open source voice assistant which “can run anywhere — on a desktop computer, inside an automobile, it even runs on a Raspberry Pi. It is open so it can be remixed, extended, improved. It can be used in anything from a science project to an enterprise software application.”
In such scenario everything becomes more unpredictable, and sales and logistics platforms like Amazon might get a bigger share of the pie. Given how much is at stake (profits, survival) we will probably see a patent war among high-tech companies to lead the future and avoid a free AI scenario.
While it’s not obvious that first-movers will have a clear advantage in AI, the ability to release AI software to an installed base of devices would give an instant advantage to those able to perform it.
Apple (IOS, macOS), Google (android) or Microsoft (Windows) could achieve this easily through an update of their operating systems, already installed on billions of devices. Amazon and Facebook could face more difficulties, as their apps are mostly run on top of someone else’s operating system. As mentioned before there is a strong incetive for companies to prevent users from installing AIs from competitors.
What cannot be denied is that any player able to secure a leading position in the AI field such as the one Google enjoys in search, would reap huge benefits across many industries.
The AI market will probably follow a power law distribution, with leader/s taking up the bulk of it. Morever, a scalable AI product recommendation system able to serve millions would exploit that capability to improve their feedback to individual users, attracting new ones in the process. This network effect means the AI service would become more attractive as additional users join.
Unless a free AI scenario prevents it, the above would lead to huge increases in valuation for those on top and, conversely, to big losses for companies in the long tail. If the AI leader is a startup the term unicorn wouldn’t do it justice and we might be having to come up with a new word. Finallly, even if its AI technology doesn’t come up on top, Amazon might also benefit of the new scenario on account of its commercial and logistic platforms.
While no one knows for sure how AI will affect our lives, some of the largest companies in the world are heavily betting on these technologies, as they expect them to bring about a lot of changes. Such changes will not happen overnight, but they will come nonetheless, and concerned companies and managers should prepare for it.
In the medium term, the appearance of AI-based personal shoppers may have a decisive impact on business operations for many companies, big on marketing and sales functions, but also on production and logistics. In a nutshell, AI has the potential to radically change business models.
Depending on the type of product, the complexity of the purchase and the amount of the transaction AI programs will get more leeway from users, from simply looking for offers to making the purchase or signing binding contracts.
The emergence of a new middleman is a challenge for manufacturers and sellers, focused on achieving customer loyalty through brand building. AI-based purchasing would generate time and cost savings, introduce rationality (absence of bias, comparability…) and could bring with it a standardization of features for products and services, with consequent effects on competition for the benefit of consumers.
However, it’s not all bad news for sellers. The most skilled will take advantage of the technology to extract more information from their customers, beat the system and generate new business opportunities: increase market share, enter new markets, customiz e their offer, etc. The ability to differentiate and adapt to a changing demand (speeding up the product life-cycle, manufacturing and logistical flexibility) will be key.
There is no doubt that the stakes are high for big tech companies. General AI is a horizontal technology, with applications across the board, so rewards for leaders can be inmense. On the other hand, the consequences of not having a significant share of this market could be devastating because of network effects and power law distributions. Search and ad models in particular will have to be rethought.
[…] what if an AI system uses anonymous browsing and hides customers’ preferences? What if the software collects information from some database instead of visiting the merchant website? AIs could delete cookies and browse from corporate servers, masking identities and repeated visits and sharing product information from websites with millions of users. There may be ways to mitigate this problem, but this is no laughing matter. In fact, shopping bots might render useless many of today’s most popular online marketing strategies.
The Internet has expanded consumer choice, and artificial intelligence-based recommendation software promises to improve the quality of user’s purchasing decisions. For brands and companies the new scenario would be an enhanced version of the not-so-old saying, “Competition is one click away”. As we’ve seen the threat is real for all kind of manufacturers and merchants, who must get ready if they do not want to end up selling commodities, or worse.
It is often said that if you are not on the internet, you don’t exist. In the future it may be not enough with being online. You will need to be on the radar of AIs to survive.
The ability of AI software to work on our behalf and push brands and apps to the background might raise suspicions among users. How to know if an AI is fairly rating products and companies? How to know if it’s the best available? Several steps could be taken to prevent this:
- Benchmark services, such as the ones used to rate processors or GPUs.
- Periodic AI competitions.
- Data standardization and use of open formats to allow data exporting. This would lower barriers to exit for users and thus work as an incentive for AI providers to be at the top of their game.
Finally, it is worth mentioning the role to be played by governments and regulators. It seems it would make sense for them to stimulate the development and adoption of AI in order to reduce cash transactions and tax fraud. However, considerations about the effect on employment effect or the strategic value of those industries affected, as well as influence from lobbies, could lead politicians to make life harder to innovators.
I could go on, but the article is already too long, so I’ll just leave some questions to the reader: how do you think artificial intelligence will affect your brand, your company, your industry? Have you already included AI in your strategy?
 Interaction time should decrease as users familiarize with the system.
 This is not a scenario a la Bostrom, where an AI gets superintelligence status in a very short period allowing it to become a singleton. That, if happening, will most probably take decades.
If you liked this post, please share it and click on the ❤ button below.