The AI fad that is fooling everyone
- 26 September, 2018 06:30
A lack of true AI hasn't stopped companies claiming to offer AI solutions where no AI is involved.
We all need technology to become simpler and smarter. It is an increasing percentage of financial spend and yet costly mistakes occur frequently.
The modern technology industry has changed and grown rapidly but very few of us operate in the high-tech world where breakthroughs are made or where new operating models are conceived. Instead, most of us work inside traditional organisations struggling to remain competitive against a backdrop of immense and unpredictable change.
One area of hope that has been touted is the advent of artificial intelligence (AI). It has been heralded as the solution to all our problems. The only issue is that it hasn't arrived. Despite numerous references to the contrary, we are still no closer to true artificial intelligence. This fact hasn't stopped the marketing machine from rebranding and realigning old solutions under the label of AI.
AI in real life
Towards the end of 2017 I found myself working in a team that was bogged down with mundane tasks like preparing minutes, action tracking and scheduling meetings. I believed that AI could help and signed-up for trials of the three most popular services available. I explored this technology for personal and professional use. I exposed friends and colleagues to scheduling AI and also searched for a transcription service that could help with the minutes.
The scheduling AI varied from good to awful. The best solution used email to convince a hotel manager that it was a real person. The worst solution almost cost me a friendship because it kept pestering for a meeting that was in the past and worse still was impossible for me to cancel. None of the scheduling services on offer could access private corporate diary systems, only public, permission based ones.
Transcription services using AI were mythical and despite reviews to the contrary, none could be found.
Successful organisations already innovate and develop using technology. They do not depend on AI in order to progress.
The experience gained over six months made me re-evaluate the current state of AI and how some choose to define it. All in all, the experience was a disappointing failure. The AI revolution, the secret productivity weapon I was hoping for, didn't materialise.
What is AI and what do we expect from it?
Artificial intelligence is defined as intelligence demonstrated by machines as opposed to humans. The term AI is more commonly applied when a machine mimics functions that humans associate with intelligence, such as learning and problem solving.
For decades, people have had their expectations of AI set by the entertainment industry. All powerful, all knowing and in some cases subservient AI representations, have captured the imagination of audiences. This creates a big gap for technology companies to fill and to date, none have come close to emulating the on-screen AI's.
The AI prize
There is a real and desperate need for AI to address our neglected problems. If the cataclysmic visions of AI are set aside, the benefits could usher in the dawn of a new golden age and a renaissance for humanity.
Instead, some organisations are investing in impostor AI solutions in a bid to simply tick the box rather than truly change and innovate. These companies are often traditional, large, dominant, but technologically less developed.
Technology usage and innovation varies from industry to industry. What seems like science fiction to one is a matter of fact to another. When comparing the technology maturity of industries, I use the analogy of making fire.
In a technologically mature industry, fire is just a matter of thought and things burst into flames. In a less mature industry, sticks are rubbed together to achieve the desired outcome.
In some of the large, least technologically developed industries, people are given bricks and told to make sparks. Despite protests that you can’t make fire this way, people are told to bash them together and see what happens. It is this last level of technological maturity that is driving the AI bandwagon along.
The AI fallacy, a harsh reality
Some organisations are perpetrating an AI fallacy on customers and competitors in order to appear innovative. However, people know that real AI doesn't yet exist. A few moments spent with any computing device is sufficient to demonstrate how basic technology is.
A lack of true AI hasn't stopped companies claiming to offer AI solutions where no AI is involved. These solutions are aimed at those who are keen to jump on the bandwagon but know little about it.
The marketers have re-purposed phrases like automation and robotics to sell modern versions of old solutions. Everything from natural language search, big data analysis, collaborative filtering and the like, are being re- branded and sold as AI. There are even distinctions made between single and multi-purpose AI solutions. The latter, as yet does not exist and the former is likely to be a collection of well known algorithms bundled together.
AI pioneers like Marvin Minsky (co-founder of the MIT AI Lab) would be saddened to see how his field of expertise has been co-opted to market and sell old technology.
The hard sell and its impact
Selling innovation to the Board is often the hardest task for a variety of reasons: past failures, cost overruns, overselling, lack of awareness, the list goes on. This is especially true in less developed companies where a lack of technical knowledge and experience makes the task much tougher. Project sponsors dress up initiatives using the AI tag in order to promote investment in technology.
Ironically, many companies that lag behind the technology curve try to develop and deliver innovation themselves. These in-house efforts often fail due to a lack of preparation and experience. This perpetuates a vicious cycle which prevents such companies from progressing beyond the less developed stage.
Is real AI involved? Ask questions about the AI components such as size of neural net and machine learning properties
This fundamental lack of understanding by the people involved ultimately sends a signal to those with the vision and prowess to disrupt established models. Two examples are the hospitality and private transportation industries. In a few short years, these two industries, which have hardly innovated at all, now face complete disruption. Both Uber and Airbnb are the culmination of radical thinking enabled by technology.
Avoid the AI trap
An example of an AI tagged solution is the so called 'chatbot'. These are often natural language search engines presented in such a way as to simulate the now familiar text chatting experience. Previously the customer had to use a search page and then select from the results. Now something similar occurs inside a chat window and is dubbed AI.
What followed were widely reported chatbot failures from the likes of Microsoft, Facebook and even Google. Some have suggested that chatbots are dead and that a lack of AI killed them.
An impromptu conversation with a well known consulting firm supported the assertion about chatbots and a lack of AI. During the conversation, I questioned the lead consultant regarding a chatbot they'd developed for a big customer. All but one of my questions went unanswered. The consultant mumbled vague and incorrect responses which caused me alarm. A partner in the firm intervened sensing the embarrassment of the situation.
Less developed companies are easily sold faux AI solutions but there are ways to avoid this:
1) Don’t run before you can walk: If your organisation is technically less developed, ask why an AI solution is being proposed. Seek proof of real world effectiveness, not just a proof of concept. Challenge the organisation's ability to understand, accept and manage the proposed solution.
2) Is real AI involved?: Ask questions about the AI components e.g. size of neural net, machine learning properties. Get advice from trusted independent technology experts.
3) Check costs and performance: AI solutions are purported to be superior but their costs are being inflated. Costs should decrease over time as the solution becomes more efficient.
4) AI is coming, watch for the signs: Perfect voice recognition is considered to be a key turning point that signals the birth of artificial intelligence. Currently most AI solutions rely on processing data from text input as it is easier and requires less computing power. Critics argue that true AI should be able to deal with inputs from any source.
Some say that all companies are now technology companies. Many feel the need to state that "we are no longer an XYZ company, we are a technology company." Not all organisations can or will make this transition and not all who want to make this transition will survive it.
The fashion for AI is accelerating the need for companies to become technology focused but not in the way that we expect. Less developed companies look deploy AI to appear innovative. Successful organisations already innovate and develop using technology. They do not depend on AI in order to progress.
The industry players in hospitality and private transportation continue to push back against Uber and Airbnb. They fight a losing battle. Similarly, less developed companies are already endangered and their extinction is just a matter of time.