The incentive to bend truth

Welcome to Beyond Human- the AI Story, where we explore the new reality where AI and humans co-exist.

This piece is written without prejudice towards any individual or company. It reflects my personal experiences on the journey I have been fortunate to live. -Mohit Sharma         

The history of technology

From the moment the first stone tools were invented, to printing press in 1440's and the subsequent penetration by information and communication technologies in the 1970's to the current era of artificial intelligence- humanity has been on a continuous spree of redefining the future through new technology. Peter Thiel in his bestseller "Zero to One" says "properly understood, any new and better way of doing things is technology."

The relatively easy access to capital through private investments and venture capitalists arising on the scene, further scaled up the pace of innovations coming through in recent decades. Enhanced and open-source access to developer tools and resources, particularly within the realm of information and communication technologies, has inspired creators and innovators to reshape the future of our world.

Growth over profitability dilemma

A side-effect of more money flowing into technology products, was the narrow focus of product companies to focus on speed and scale, rather than real value and impact for the client and the product itself. There were numerous phrases like "blitzscaling" propelled by legendary innovators, who professed about 'speed over quality', and 'growth over profitability' as a mantra for new innovators and founders.

The message reached far and wide due to social media and the quest of new founders, looking for meaningful guidance to get funded. Template-led investors, also heeded to the call from many such founders. The common hypothesis that only 1 out of 100 of your investments as an investment team will be extremely successful and it will balance out the other 99 which were a failure, also added to the hype.

But in the real world, you can sell the products on hype- but you cannot make it work on hype alone. Template-driven investors, what Elon Musk calls as 'bean-collectors' and fresh breed of young graduates out there with the intent to 'build the next unicorn' added more speed to the phenomena.

Finding the 'Problem-Solution Fit'

There is no bigger disappointment than losing on an investment, whether you are a buyer at a large enterprise, who invested in the next big-thing, which fails to deliver and capture value or you are an angel or a VC which backed an innovative founder and the product fails to be profitable and effective.

There are thousands of illustrative case-studies on product websites trying to impress prospects about the benefits a product can potentially deliver.

There is no common certifying authority about whether this is a validated information on display on the website and real impact, or just a little bit of reality was bended in order to keep the investors happy. A human's natural tendency to trust another human, is leveraged to drive growth over profitability.

Definitely, the incentives to bend the truth are much bigger and lucrative than the risks of staying real.

Understanding the Problem Solution fit

In a world swamped with influencers, and virtue signalling, in the middle of 'AI era'- it is extremely difficult for buyers of technology to take decisions based on merit. The effect of AI-washing is real.

Private companies which sell products are under no compulsion to make their financials public and hence it is not possible to understand how much of the growth they achieved so far, came from genuine customer centric efforts. I do not mean to say all of the private businesses are marketing led, but a large chunk of the technology sales that happen today are accelerated completely through sales and marketing.

They are not bought, they are sold. The sellers are in a rush, just like the buyers.

So this trend creates so called number one solutions in their space, in business for last 16 years and still not profitable.

Does it make sense to you? It doesn't make sense to me. Even if I was selling candies to kids, I will have to have profitability in place driven by customer demand, for my business to stay afloat.

In the recent past with even the pen-seller in a bus using AI to position his/her product, I have seen a lot of marketing reaching out to me with taglines like ' We have been doing AI, when nobody even knew about AI'.

Great marketing tagline dear, but AI as a term exists since 1953- its the technology maturity which has been a recent development. Good enough to grab eyeballs, not justified enough for the customer to buy.

The effect of purchasing decisions taken due to such marketing ( and mind it -this includes some key decision makers being offered advisory board positions across these same product companies) is that we have a lot of such sophisticated technology solutions sold and deployed in production- for which the business stakeholder is asking, "Can you please help me quanity the benefit of this implementation?", even after 2-3 years of runtime.

If the product you invested in has not been able to capture and deliver value to you in 2-3 years, either the marketing was a blatant lie, including the case credentials you were shown or you ended up choosing the wrong solution for the right problem.

And there is a 9 out of 10 chance of the buyer falling victim to the bending of truth which the product sellers leveraged.

The Solution: the Intelligence continuum

In order to solve problems for some of the clients that I work with and help them understand the context around the problem they are dealing with and how to find the right fit, I have used this simple continuum as a framework, which has worked everytime in finding the right fit solution.

If you are a buyer of technology or some of the 'AI washerman' are reaching out to you with a lot of push, feel free to use this to map your problem to find the right solution.

One simple rule is, if your process is not even standardised, first you need to simplify and standardize your process, before even planning to put the '#1 AI solution in the space'. This is a continuous gradient, on which intelligence and autonomy will move from left to right.

Here is a brief description about the different stages on this continuous, with their attributes:

1. Manual: the process steps are highly manual, with no clearly defined rules. The process participants might have developed a 'tribal knowledge', an instinct to know what to do next over a period of time. This knowledge is people driven and innate in the participants. For example, in the old taxi system, the drivers knew all the different routes and had some leading practice ways to make the rides efficient and effective. Not all drivers were equal in skill and competence.
2. Standard: the process steps are still relying on manual input, but the roles and responsibilities of each participant are clearly defined with each participant playing to the rules. There are systems of governance in place which ensure, anyone breaking the rule and leading to value leakage or delays, will be spotted and appropriately dealt with. In the taxi example, this is the time associations, governing bodies and unions started regulating the rules around boarding points, booking etc.
3. Automation: when the rules mature, over a period of time- a lot of repititive activities make themselves visible. This is the time to put some point solutions to automate them. RPA bots used by back-office are a classic use case. The goal is mostly to reduce manual effort across repetitive activities and these point solutions are an easy pick for such things. For the taxi system example- The logging, availability of vehicles through wireless radio, is the stage. This helped optimize the availability and revenue further.
4. Orchestration: over a period of time, these bots span a lot of process touch points and it becomes extremely complex to understand where a human hand is there and where a bot plays the role. This is a moment to have orchestration, a human and machine partnership emerging where an overarching platform can become the common engine to orchestrate, set in sequence, monitor and deliver the service. In the taxi example, this is the moment- the ride-sharing system through apps like Uber have arrived.
5. Intelligent Automation: this is where the orchestration system has set in motion so much visibility across a complex process, that the business can take a call to increase autonomy across certain segments of the business and leave it completely to machines. This stage needs some enhancement in capabilities through technology for some independent segments. In the taxi example, integration with GPS made monitoring of rides, and finding the right driver easier. The entire system got a boost of intelligence and autonomy.
6. AI Driven: this is a stage, where the intelligent automation can span across the entire process using a ubiquitious human-like intelligence to take decisions, and execute the same. AI will need data, loads of it- if you have not passed through the stages 1 to 5, there might be gaps in data and the machine learning algorithms might not be able to fix those gaps on its own, since data is the fuel on which it runs. If you invested in an AI product, and are struggling to determine the value capture, just check if what you originally needed was a stage 1 problem and you are putting a stage 6 solution which requires a different level of maturity across the architecture and landscape. In the taxi example, you can relate with dynamic pricing algorithm, ETA and route management, destination prediction and search ranking, driver ride matching, personalized marketing etc. Was it possible to deploy all these in stage 2? Many product companies sell their stage 2 solution, disguised as a stage 5 solution to a customer at stage 2. By now, you know what I mean.
7. Fully Autonomous: this is the ultimate north-star, the elixir or the stage which drives emotions of excitement and fear both. This is the fully-autonomous machine intelligence. In the taxi example, the autonomous air-taxi system expected to be launched in Dubai, or the kind of self-driving 'ride as a service' pilots running in other parts of the world will help illustrate the point. This comes once AI has matured enough, and has demonstrated signficant maturity, business value and autonomous run over a couple of years.

Conclusion

There is a right product for every problem and no two products deliver the same value at the same stage. One can easily start from the problem context, and then explore solutions which are best fit for each stage of the intelligence continuum. It is a descriptive, not a prescriptive framework. What this means, the user is free to use it the way they want, there is just one rule- you cannot put a Stage 6 solution to address a stage 2 problem.

The solution for stage 2 are different solutions and if you make the right selection, within weeks you will start seeing the business value. Ofcourse there will be some calibrations required- but you would never have to wait years to see the business value delivered.

If the ultra-modern AI solution you deployed is not giving you benefits, ask yourself two questions:

• What was the problem you wanted to solve? Was AI the right solution?
• What is the solution you bought? Is it AI in reality? Or just a transaction level 'if-else' formula projected to you as AI as it saves some seconds of work for a process participant?

Tell me what you think about this. Feel free to share it with others, you know of still waiting for the miracle to happen and their frog to transform into a handsome prince.