The Agile 2016 conference annual Industry Panel Discussion session examined Agile Trends and Future Directions. The panel discussed the implications of artificial intelligence and machine learning on the software industry, safety and ethics in AI systems, whether the Agile Manifesto should be updated, and the impact of tools on agile adoptions.
The panel consisted of Tom Grant, Ray Arell, Steve Denning and Rebecca Parsons. It was moderated by Jim Newkirk.
The first question focused on the impact of deep learning and artificial intelligence on the agile movement.
The panel discussed how deep learning algorithms are already in use in many aspects and industries - the example was given of Spotify's Discover Weekly capability which learns the subscriber's taste in music and presents new music which listeners find "uncannily accurate". This has not resulted in the need for less software, but rather a different approach and probably more development; there will be a significant change in how the programming is done - there will be more and more teaching of AI's instead of writing lines of code. The challenges will remain the same - getting the human interaction aspects correct and the skills needed today for getting those aspects will remain the same, while the implementation approach will probably change in ways we don't yet know.
A challenge in the coming complex environment will be the testing of the deep learning algorithms - how will we be able to have confidence that the results are accurate, unbiased and safe, for example in drug design? This will require new ways of examining logic paths and testing approaches which we need to start thinking about now.
The next question looked at ethics and safety in AI systems, using the example of the recent Tesla crash. What is the bar for success in machine learning systems? In 2014, there were 32,675 driver fatalities in the US, yet the reaction to a single fatality attributed to an AI seems to have caused a significant backlash against the technology. Is zero tolerance really necessary?
The comparison was made between the Tesla approach to development and that of Boeing in their flight control systems. In the FAA regulated environment the programmer, manager and QA person are all required to sign and prove that the software has been built to a rigorous standard that has been thoroughly quality assured and quality controlled. This sort of rigor needs to be applied to other areas where software algorithms have the potential to cause harm or death. This is much more of a societal than a technological question and we need to educate society to understand the implications of these types of technology.
The next question looked at the Agile Manifesto - should it be changed to bring it up-to-date given that it is now 15 years since the Manifesto was written?
The Manifesto is a historic document - it was right for the time it was written. The sentiments and philosophy behind the Manifesto are as valid today as they were in 2001. The focus on software was appropriate then and is less so today as agile approaches are being used outside of software, but that doesn't mean we should redraft the Manifesto. It is reasonable that when discussing the Manifesto, it should be discussed in terms of what the ideas mean today - the key is that the values and principles guide what it means to be agile while allowing adaptation of the words to the context in which people are working. Don't change the manifesto but feel free to produce representations which are context-specific and resonate with the values and principles of the original Manifesto. Avoid zealotry and bigotry and embrace the agile approach of continuously learning. What "agile" means is being continually reinterpreted by the community, it's not about a document - it's about the dynamism and adaptation of the ways of working inspired by the ideas embodied in the Manifesto.
A problem that is seen is the adoption of agile practices and techniques without understanding the ideas, the values and principles of the Manifesto. These adoptions are often challenged, the teams are able to follow the rites and practices, but without the foundation of the values and principles the agile mindset is missing, and the teams and organisations don't achieve the humanistic, sustainable culture improvements that are possible when the values and principles are at the foundation.
Agile adoptions are often measured on the practices and processes teams abide by, rather than the hard-to-measure aspects such as team collaboration, engagement, trust and value delivered.
The commercial pressure to sell tools and processes, on recipe-based adoption of defined frameworks and rulebooks is damaging the reputation and removing value from the Agile movement. Unless the emphasis moves away from tools and processes and back to individuals and interactions, to teams and teamwork in order to the focus on people and sustainability, then agile is doomed to go down in history as yet another process improvement approach which had little impact, following on the management movements of the 20th century such that started with noble aspirations and became corrupted into bureaucratic monsters.
The final question was what is the biggest barrier to Agile adoption that you see in your work?
- We have not adequately communicated the value to the executives and other portions of the organisation. We need to speak their language (finance, marketing, HR) rather than expecting them to learn our language.
- Use the ideas from organisation design and align the organisation as a whole around a common set of values, use the language and terminology of the wider organisation, and craft a new mindset that is agile at the core- it doesn't make sense to have one set of values for IT and a different set for the rest of the business.
- Lack of strategic alignment- organisations often don't articulate the goal for agile adoption in business terms. It's not about adopting an approach or a process in IT, it's about achieving measurable outcomes for the organisation, such as increased customer satisfaction, improved time to market, and reduced cost of development. These need to be clearly articulated.
- The individuals who believed that everything was OK before, and are therefore uncomfortable with the acknowledged ambiguity of an agile project. The myth that development was predictable and simple and that if only you planned better you would get better outcomes. That world may have existed in the 1960's but it hasn't existed since then, but the belief in predictability still persists. We need to change these expectations.
- The perception that success is about extracting value from the organisation and returning it to shareholders, rather than generating value through customer and employee engagement. Organisations which are able to shift their focus do in fact result in higher shareholder returns as a by-product of customer delight and employee engagement.