Everyone agrees there’s great value to strategic planning. In businesses of any scale, it’s essential not only for ensuring survival, but also for progress towards goals, and keeping pace with competition.
In a recent Forbes article, a survey cites that, unsurprisingly, 90% of leaders believe in the necessity of such planning, with most also agreeing it should look three or more years into the future. (Far fewer believe their companies succeed at these goals, and everyone the world over agrees it’s far easier to plan than actually implement, for obvious reasons.)
Now factor in AI.
Since the emergence of ChatGPT in November, 2022, we’ve witnessed a steady stream of evolving AI technologies, some hype, but much of it real. The plateau for large language model development has risen at least twice since then.
[Check out The PTP Report for a look at how open-source models reached the latest plateau earlier this year, with additional coverage in our bi-monthly AI roundup.]
We’re still discovering what these products can and can’t do well.
For businesses that desperately need good planning but already struggle to implement, how can they incorporate such a fast-changing, and ill-defined product?
One thing is certain: we can’t know for sure what AI will and won’t be capable of doing two years from now, but we can be certain it will look quite different than it does today.
It may be nearly as foolish to embrace today’s AI as a finished product as to wait and see what the future holds before acting at all.
This is the goal of today’s article: considering how best to navigate such a rare blend of turbulence and opportunity.
The Current State of AI Integration
I am a big believer in the transformative power of generative AI, and as I wrote in a prior entry, 97% of business owners agree with me, saying ChatGPT (specifically) will help their businesses. 72%+ had already begun adopting it as of May, with 85% having no AI policies in place at the time.
The growth curve has been steep, rising in plateaus to date, with OpenAI’s large language model GPT-3.5 rolling out in 2022 (powering ChatGPT which launched in November of that year), followed by GPT-4 (the current plateau for AI sophistication) replacing it, in March of 2023.
GPT-4o (a far smaller upgrade) appeared in May, 2024, with OpenAI acknowledging both that this was not a full step, and that the next one is well underway.
Open-source solutions have lagged perhaps a year behind, though now Meta’s Llama 3.1 and others bring GPT-4-level sophistication to a broader market.
From here there is all manner of speculation, including that OpenAI’s Strawberry/Q* will come soon and build on Google AlphaProof’s capacities for reasoning and research, that the end of public data at scale may signal a move to private, corporate data or synthetic data, that larger context windows and greater autonomy will unlock more agency—but again, regardless of all, we can be certain that change is coming. And soon.
Many companies have leapt at the chance to automate tasks and improve customer service and data analysis with these technologies, but it’s still an exploration, R&D at scale, rather than pragmatic tooling, and without comprehensive, organization-wide AI integration strategies, it can prove short-lived and ineffective.
Problems and Solutions
AI can help companies improve efficiency and personalization both. It can augment the speed and scale of data-driven decisions. But data safety, ethics, and cost are all critical challenges of AI integration.
Let’s look at what I see as some of the biggest problems in this arena and potential solutions.
Problem: This Pace of Change
I’ve spent much of the article so far considering why the speed of AI rollouts make it difficult for companies to keep up. The systems we use today may well be outdated by the time they’re truly integrated into your workflows.
Solution: Flexible AI Adoption Strategies
Companies must be nimble, ready to adopt new advancements such as coming models with larger context windows and real-time decision-making capabilities.
Stay flexible in cloud services commitments and implement modular systems that can be easily updated or upgraded. The more flexible your infrastructure the faster you can incorporate new tools without having to start from scratch. To this end, consider change as more continuous than fitful.
Problem: Data Quality in AI Integration
Garbage-in, garbage-out. It’s true with data in all applications, and certainly the case with generative AI, which needs high-quality, accessible data to thrive and reach its potential. The more fragmented, irregular, or incomplete your data sources, the harder, and lower quality, your AI implementation will be.
Solution: Invest in Data Integrity and Compliance
To get the most from AI solutions (now and forever) you must have your data in order. This means clean, standardized, and well-prepared, with refinement and oversight incorporated from the start. Apply strict access controls, optimize your structures and architectures, and to the extent possible, break down silos that prevent aggregation and insights at scale.
Easier said than done, I know. But take advantage of experts that are here to help!
Problem: Runaway Price and ROI
2024 was supposed to be the year we start seeing returns on AI investments, but the picture is still quite muddy and hard to quantify with much objectivity. Without being clear on the costs of AI integration, how do you go about managing AI costs to measure your return on investment? Without clear numbers, it can be difficult to justify or even budget effectively.
Solution: Iterate to Grow
As we’re in an R&D phase now, I say start small as necessary, but start now. Implement AI in tried areas where companies are seeing impact, such as in customer service and streamlined data entry. Reward your teams for AI experimentation in your business with challenges and contests, which not only get the conversation started, but can also give quick wins that demonstrate practical value.
Problem: Employee Concerns and Training for AI
When ChatGPT exploded on the scene the news centered on layoffs and how careers would become vulnerable. And while AI use has grown exponentially since then, worker concern remains high. Employees may resist workflow changes, being threatened by the technology, or simply because they are unfamiliar with it.
Solution: Continuous Learning
I mentioned above (and in a prior article) that a great way to get started is by launching a contest, for example, for creating their own GPTs, for example, that could improve workflows. Of course this must be accompanied by smart policy, including at least a rudimentary education on data privacy. But by getting workers hands-on experience ASAP, they begin adapting to AI advancements themselves, encouraging employees to share what they know instead of keeping it secret for fear of repercussions.
This is only a start, of course, as regular training will be necessary to keep up with the pace of change, but it is certainly a step in the right direction.
Tackling Unpredictable Evolution
The big tech companies are following their own AI strategic planning, and to a one this includes the development of Artificial General Intelligence (AGI).
Opinions remain wildly divided about when this will be and how it will impact businesses, but it is a clear goal consuming billions in investment dollars.
Any movement in this direction will have a profound impact on work (and society as a whole), making it foolish to completely disregard out of fear, skepticism, or ignorance.
To make your company ready for any future, I recommend staying as flexible as possible with your own AI implementation roadmap. Considering multiple possible futures is simply a must for future-proofing AI investments, and to the extent possible, I encourage creating variable visions for the future of your company (with and without AGI, for example, over the nearer term).
Regardless of what ultimately happens, by having considered these possibilities, you will keep your business from being caught flat-footed.
Conclusion
Strategic planning is both necessary and an exercise in pure speculation, leading many employees to wonder how much value it has for practical ends.
This has always been true and maybe even truer in this age of rapid technological change.
But while the temptation is great to shrug off the hype and keep focus solely on the now, prudent leaders must consider the possibility of significant, substantial change being nearer than it may be conventional to accept.
AI is evolving quickly, and businesses that wait for certainty may wake one day to find themselves suddenly several steps behind their competitors.
I recommend staying as flexible as possible. Start small, stay agile, and experiment with new innovations now, and don’t skimp on your planning, however tempting, and for multiple possibilities.
