Introduction
This paper describes and explains the concepts and terminology behind what is today being termed as Artificial Intelligence. Further, it describes how these concepts relate to the field of project management offering opportunity for better, more effective project planning and control.
What Exactly is Artificial Intelligence?
There are many definitions of Artificial Intelligence or AI. In fact, a Google search today returns 1.18 billion results. One of the funniest definitions I have run across is “AI is whatever hasn’t been done yet” – now there’s a vague and unhelpful answer!
One of the more useful definitions I have found is “AI is the ability of a computer program or a machine to think and learn. In general use, the term ‘artificial intelligence’ means a machine which mimics human cognition.”
So, machines being able to think and learn seems to be the crux of AI…
The way humans think is through what is called cognition (stems from the Latin for “know” or “recognize”). It is the scientific word for a thought process i.e. the mental action of acquiring knowledge and understanding through thought and experience.
The way humans learn is through either observational or associative means. Observational learning is watching others behavior e.g. watching your parent drive a car. You learn from watching which levers and switches they push as they drive along. Associative learning, on the other hand, is learning by establishing connections between events. You know there will be thunder is you see a lightning strike.
Humans make decisions based on thought and learning. We make sound or good decisions based on observational reasoning as well as associative patterns. We also sometimes make bad decisions that we can learn from to make us smarter the next time around. So our thought process gets smarter, the more we learn.
If a machine can acquire knowledge and understand or recognize it, then it too can start to make informed (and hopefully good) decisions for me. I believe AI is really all about a machine being able to make an informed decision that is a sound one. It is a decision support system (DSS) that helps me make a better decision faster than I could have otherwise made.
The Problem with Project Planning Today
One of the hardest challenges in project management is accurately forecasting future outcomes (project completion date, total cost) of very complicated and highly uncertain endeavors (projects) – we call this planning.
As an industry, we have developed some pretty well tried and trusted techniques such as Critical Path Method (CPM) to try and help model project outcomes but these models are only as good as the inputs we feed into them. Any worthy planning tool today uses CPM as it’s underlying forecasting engine but as the planner we are still left with the onerous task of knowing not only which activities to include in our plan, but worse, what should their durations, cost and even sequence be? CPM does little more than convert durations and sequences of durations into a series of dates. It doesn’t help one bit with:
- What scope should I focus on when building my plan?
- What activities should we include?
- What should our durations be?
- What is the true sequence and logic between our activities?
If CPM were the be-all and end-all solution, then we wouldn’t continue to experience project cost and schedule overruns. The problem isn’t CPM though. The problem is our inability to accurately model what we think will happen during project execution because of a) the huge number of variables (tasks and sequence) and b) the huge number of uncertainties associated with those variables (duration or scope uncertainty).
Schedule risk analysis tools help tell us how bad our forecast may be but they do nothing in terms of telling us what the inputs to our schedule should have been in the first place.
This is why I believe AI can massively help project planning. If AI can help the planner by making suggestions that are sound, then the immense challenge described above starts to become surmountable. Added to that, if our planning tool can also start to make better suggestions by observational or associative learning, then we are headed down a seriously valuable and exciting path…
AI Categories
If you thought the number of AI definitions was bad enough, Google returns 754,000 hits when searching for Types of AI and sadly, very few of them return a common set of type definitions.
For me the best way to categorize the types of AI is by categorizing as follows:
ANI
Artificial Narrow Intelligence (ANI) also referred to as Weak or Applied AI, Artificial Narrow Intelligence is AI that specializes in one area. Examples of this would be the IBM Deep Blue computer beating a chess master at a game of chess. The machine was programmed to be very good at one thing – playing chess. Believe it or not but Apple’s Siri is also an example of ANI. She is programmed to respond to a limited set of questions but go beyond those questions and she cannot give an informed answer. The majority of today’s AI solutions are ANI-based. My thoughts and discussion on AI and project planning later in this paper all focus on ANI AI.
AGI
Artificial General Intelligence (AGI) is also referred to as Strong AI or Human AI. AGI refers to a computer that is as smart as a human across multiple domains. Computer science is nowhere close to achieving AGI yet.
ASI
Artificial Superintelligence (ASI) superintelligence is “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills.” Even further afar than AGI, let’s just move on!
Current Approaches to ANI AI
Today, our implementation of ANI AI can be loosely classified into two categories: expert systems and neural networks.
Expert (Knowledge-Based) Systems
Originally developed for use in the 1980’s, expert (or knowledge-based) systems (ES) really came into their own as computing power got strong enough in the 1990s’. An expert system is a program running on a computer that uses a set of rules to answer a question (typically in the form of IF…THEN).
When asked a question, an ES will filter a set of data, based on rules, to establish a sub-set of what it believes is the answer. As a general rule, the more rules that can be used to answer the question, the stronger the chance that a correct answer will be given. When determining what type of animal, simply running “IF number_of_legs = 2” doesn’t narrow down our search enough to give us a useful answer given the large number of animals with 2 legs. String this to an additional set of questions relating to height, weight, habitat, pouch etc. and we can quickly deduce a reasonable answer.
An expert system comprises a knowledge base and inference engine. For a project planning tool, the knowledge Base may contain data pertaining to activities and their durations for different types of project. The inference engine is then responsible for trying to return a sub-set of this knowledge base back to the planner based on the question they may ask e.g. what activities should I include for my engineering scope of my hospital project?
In this example, what is further useful is to understand how confident is the computer that the returned suggestion is correct. This is where the likes of fuzzy logic come into play. Rather than returning a definitive list of activities, the AI engine should return a sub-set of activities with associated degrees of confidence about their relevance.
Neural Networks
A neural network (NN) tries to simulate the way a brain processes, learns and remembers information. A NN learns from experience. A NN looks for similarities in information that it is fed and previous data and then makes a decision based on that – it looks for patterns. This pattern matching is based on what is called machine learning – you have a teach a NN what is a match and what is not. Feed in enough examples of characteristics of an alive human (breathing, pulse, eye movement) and a neural network will start to establish a pattern as to whether those inputs drive towards a correct diagnosis of alive or dead?
There are various forms of machine learning in a neural network including:
- Supervised: e.g. feed in an activity that has zero total float and tell the NN that the activity is ‘on the critical path’. After feeding in enough of these activities, it will establish a pattern that matches zero float activities to critical path activities.
- Unsupervised: feed in activities but don’t tell the NN which are on the critical path or not and let the NN try and categorize the activities based on it’s various attributes e.g. total float. In this instance, the NN will perhaps group into zero and non-zero float without knowing this relates to critical path – it simply groups activities together.
- Reinforcement: the teaching through reward. e.g. teaching a kid good behavior by offering a treat. With regards to our planning software examples, perhaps the license cost of the software should automatically go up or down depending on how good the AI engine suggestions are!!!!
Which AI Approach is Best for Helping with Project Planning?
Unlike neural networks, expert systems do not require up-front learning nor do they necessarily require large amounts of data to be effective. Yes, expert systems can and do absolutely learn and get smarter over time (by adjusting or adding rules in the inference engine) but they have the benefit of not needing to be ‘trained up front’ in order to function correctly.
Capturing planning knowledge can be a daunting task and arguably very specific and unique to individual organizations. If all organizations planned using the same knowledge e.g. standard sub-nets, then we could simply put our heads together as an industry and establish a global ‘planning bible’ from which we could all subscribe. This of course isn’t the case and so for a neural network to be effective in helping us in project planning, we would need to mine a lot of data that, even if we could get our hands on, wouldn’t be consistent enough to actually help with pattern recognition.
Neural networks have been described as black boxes – you feed in inputs, they establish algorithms based on learned patterns and then spit out an answer. The problem is, they don’t tell you why because neural networks don’t understand context! I honestly don’t think that as a diligent planning community, we should rely on a system that doesn’t have understanding or even worse, cannot explain why a tool can come to a given answer. “I’m pretty certain you need these activities just because I know” is not as useful as “you need these activities based on previous projects X, Y, Z and your currently defined scope and the phase of the project you are currently in”.
Expert systems tend to excel in environments that are more sequential, logical and can be ‘tamed’ by rules – doesn’t that remind you of a CPM network? Neural networks pertain more to problems such as recognition through pictures e.g. project drawings and BIM.
Where I do believe a neural network approach is useful in a planning tool is in the making the tool smarter. As mentioned, expert systems can get smarter but they need to be trained. If we can track a planner’s reaction to suggestions made by our expert system, then those reactions can be used to potentially adjust the weights we give to the various attributes in our expert system, for example.
So then relating back to our original definition of AI, for a project planning tool, use an expert system to think and use a neural network to learn. Combine these two and we have at our disposal, an incredibly powerful planning aid. That is exactly what we have developed at BASIS.
Should We Embrace or Avoid AI in Planning?
I don’t think we really have a choice and for good reason. AI is not a magical, mysterious, be-all, futuristic solution– it’s already here. It is simply an incremental step in leveraging computing power to make our lives easier and better. If we can apply that to the science of building project plans, then of course that has to be a good thing.
Where I would suggest caution is to those who believe you can “push the AI button and automatically spit out a project schedule”. That is simply not possible and in reality, not helpful. As planners, our expertize and value is as much in the input and reasoning behind the building blocks that go into a plan as the outputted plan itself. I like to call that HI or Human Intelligence. AI makes that process better – plain and simple.
Mix a little AI and HI and I think we are going down a very progressive path…
Dr. Dan Patterson, PMP is the founder of BASIS, a project management software company focusing on knowledge-driven planning. He is a renowned thought leader in the project management industry developing numerous tools that are commonplace in the project controls arena today.
More can be found at www.basisplanning.com [1]. Dan can be contacted at dpatterson@basisplanning.com [2]