According to Eric Brethenoux, Distinguished VP Analyst at Gartner, AI is more than just generative techniques, it's a form of augmented intelligence. Properly understood, AI is really a set of computer engineering techniques that solve business problems.
Brethenoux says there are three ways companies can approach AI, to stay within boundaries, to push them, and to break them. The specifics of a company’s AI investment portfolio varies from industry to industry, but what’s important is how they think about the composition of that portfolio.
Here’s a breakdown of those three approaches and something to think about for anyone in any company considering their AI investments.
1. Within Boundaries
80% of initiatives of companies' AI portfolios are here. This level focuses on productivity gains in established ways of working. For example, as Peter Lee, Corporate Vice President of Microsoft Research and Incubations, says, doctors have begun using AI to write letters to parents concerning the state of their children’s health. With the right prompt, AI can write a letter that’s clear, factual, and empathetic, emphasizing the role that both doctor and parents play in restoring a child to optimal health.
2. Pushing Boundaries
17% of initiatives are here. In this case, AI helps to show new best practices in established ways of working, which is another level of change from doing things the old way, but faster.
An example of this comes to us from the Japanese game of Go. Go masters are now learning new ways to play the game from a series of evolving Go AI systems, from deepmind's AlphaGo being first AI system to beat a professional Go player in 2016, to KataGo beating AlphaGo, to Far.AI finding a strategy to beat KataGo and other AI systems in 2022.
Adam Gleave, CEO of Far.AI, says AI has upended what were thought for centuries to be the game’s best practices. These systems have left leading Go players behind. Lee Sedol, a leading Go player, has even said AI systems are unbeatable.
Despite the awe-inspiring record, AI isn’t quite perfect. AI has a very good average case success rate, but it fails badly and abruptly when faced with the worst possible cases. This points to the need for AI to aid humans, rather than replace them, so that humans can regularly evaluate those AI systems whenever needed.
3. Breaking Boundaries
Only 3% of initiatives are at this level, where AI can be said to have agency.
Natural products are the source of a third of all medicine, though only 1% of all natural products are known to science. Enveda Biosciences is using machine learning to analyze the chemical structures of biological samples, and then understand how similar these samples are to existing drugs and how available they are.
This, says Tom Butler, VP of Machine Learning at Enveda, aids the drug discovery process and helps drug-hunters prioritize.
Lessons Learned For Your Portfolio And Future
Companies will have to order their portfolio according to what’s proven, what’s possible, and what they aspire to.
AI use cases with clear implications for the competitive landscape are what every company is investing in. The next step is what seems aspirational, but is doable with current technology. The last step, the riskiest and at the same time the most necessary, is to invest in developing trends, where the possibilities are speculative and the rewards are a sizable advantage over the competition.
Employees need to be AI literate, and ERPsim is a great way for them to learn. With its conversational AI, predictive decision-making and real-time integrations, ERPsim is a business simulation game that helps employees understand the power of AI in their day-to-day work.
Brethenoux says the risk of playing it safe has never been higher. Looking at these examples, and the lessons I hope you’ll take from this, I can’t help but agree. Think of how this applies to your own industry, what your AI investments are and, most importantly, what they could be with ERPsim.
Comments