if you're waiting for ai to slow down so you can catch up, i think you're making a mistake.
there will always be another model, another tool, another announcement. waiting until you understand all of it gives you an excuse to stay where you are. meanwhile, someone else is learning how to apply it to the same work you do.
my view is straightforward: learning how to work with ai is becoming part of the job. for businesses that want to remain competitive, treating that learning as optional is a decision with consequences.
you can dislike the hype. you can question the companies building it. you can be concerned about what it means for your profession. i have questions too. those concerns deserve involvement from people who understand the work being changed.
stop letting fear make the entire decision for you. start asking where ai could help you do something better.
the pace matters, but so does what is coming together. language, code, images, audio, video, search, and predictive systems can become parts of the same workflow. a change in one area creates possibilities in another. a better way to write software can make a creative tool easier to build. that tool can give someone with an idea a way to make something they previously couldn't afford to produce.
the gap i worry about is practical experience. someone who starts with a small project today gets to discover what works, what fails, and what needs their judgment. they carry that knowledge into the next project. after enough repetitions, they have a way of working that another person cannot pick up by reading a list of prompts.
if keeping up feels hard now, postponing that experience will make it harder. i expect the capabilities and connections between these systems to keep developing. you need a habit of learning that can move with them.
there is already plenty of activity to learn from. stanford's 2026 ai index reports that 88% of surveyed organizations used ai in 2025. that measures adoption within the survey, not proof that every deployment delivered value. still, it is a clear indication that businesses are putting these tools to work. stanford ai index: economy.
the world economic forum's 2025 employer survey puts ai and big data among the skills expected to grow fastest in importance through 2030. curiosity, creative thinking, and continued learning also feature prominently. these are employer expectations, but they point toward a practical combination: learn the technology and keep developing the judgment to use it. the future of jobs report: skills outlook.
i expect ai to become more deeply involved in daily work and personal life than most people currently imagine. potentially even more deeply than the internet is today. that is my forecast, and the reason i'm putting time into this field.
think about how little you notice the internet while using it. you buy something, find a route, send money, watch a film, or work with someone in another country. the connection sits inside the activity. i expect ai to follow a similar path, becoming part of the tools and services around us until using it rarely feels like a separate event.
that future gives us a lot of decisions to make. getting involved now gives us more knowledge to bring to them.
software engineering is an obvious place to start.
if you write software, take a problem you understand and work through it with an ai tool. ask it to explain an unfamiliar part of a codebase. explore an implementation. have it propose failure cases you might have missed. then inspect the result, run the relevant checks, and see whether it helped.
you still need to understand the system. generated code can contain a subtle defect just as easily as handwritten code. your ability to recognize that defect, explain the tradeoff, and choose a better approach is part of what you bring to the work.
i care about getting more room to solve the hard parts. if a tool helps with repetitive implementation or exploring alternatives, i want to find out how useful it can be. i'll judge it by the finished work and the effort needed to get there.
the same opportunity exists in creative work. a musician could explore an arrangement or use a generated reference to communicate a direction. an illustrator could experiment with a composition before committing to the final drawing. someone making a video could develop a storyboard, compare visual treatments, or prepare a rough concept for collaborators.
there is room to use these tools while caring deeply about craft. your taste determines what belongs. your experience tells you when an image feels empty, a musical idea goes nowhere, or a scene misses the point. decide what role you want ai to have in your process, and pay attention to permissions and the terms attached to material you use or publish.
for an author, the useful starting point might be feedback on structure. ask where an argument loses its thread, which passage repeats something, or what a reader might misunderstand. use that feedback to make your own decisions. a publisher could evaluate help with metadata, descriptions, or production tasks. the responsibility for the finished publication still belongs to the people putting their names on it.
there is evidence that creative assistance can help. a 2024 experiment published in science advances found that access to ai story ideas improved readers' evaluations of short stories, especially for participants with lower measured baseline creativity. the stories also became more similar to one another. that makes a good case for exploring the assistance while continuing to develop your own ideas and voice. doshi and hauser's study on ai and creativity.
i want more people to be able to finish the things they care about. the book they keep putting off. the song they can hear but haven't arranged. the short film that has lived in their head for years. experimenting with a tool might help them get past one of the obstacles. that is worth trying.
education and personal growth deserve the same attention. you can ask for another explanation, practice a conversation in another language, or work through questions about a subject you have always wanted to understand. a teacher can explore lesson ideas and exercises, then adapt them to the actual students in the room.
the goal should be learning you can carry away from the interaction. ask for a hint, try the problem yourself, explain your reasoning, and check it. the oecd's 2026 digital education outlook describes the promise of generative ai when it is guided by sound teaching principles. simply getting an answer produced is a different outcome from learning how to reach it. oecd digital education outlook 2026.
in trading and financial research, i'd start with organizing source material, examining data, or building a way to investigate a hypothesis. the calculations need to be verified, and a convincing explanation of a trade does not establish an edge. faster research gives you more opportunity to examine an idea. it does not guarantee profits. finra's guidance on generative ai discusses potential applications alongside firms' continuing responsibilities for how they use the technology.
these examples require different tools and different levels of review. they also offer plenty of places to begin without handing over an entire business process.
for a business owner, i would start with something employees already find frustrating. perhaps they spend too much time locating information across approved documents. perhaps incoming requests need sorting before anyone can act on them. perhaps preparing a useful first draft of a proposal consumes hours that could go into understanding the customer.
pick one of those problems. define what improvement would look like. then find out whether ai helps.
the study generative ai at work gives a concrete example. researchers studying a customer support deployment found that ai assistance increased issues resolved per hour by 15% on average, with different effects across workers. the agents could edit or ignore the suggestions. that result belongs to the setting studied, but it demonstrates why a business should investigate assistance on real work and measure what happens.
some problems call for an llm. others need a predictive model, ordinary automation, or clearer rules. in when the answer has consequences, i explain why i care about auditable computations and evidence of accuracy. that discipline belongs inside an ai adoption effort. enthusiasm should make us willing to experiment and serious about checking the result.
here is how i would begin this month.
- choose one recurring task you understand well enough to judge. keep it small enough that you can review the whole result.
- record how you handle it today. include the time, quality, and common problems so you have something useful to compare against.
- try an appropriate tool with data you are permitted to use. give it the context and constraints the task actually needs.
- check the finished result. include correction time, missed details, and tool costs when deciding whether it improved the work.
- keep what helped, change what didn't, and share what you learned. then try the next task.
each attempt should leave you with something you can use in the next one:
flowchart TD
a["choose one useful task"] --> b["record how you do it today"]
b --> c["try an appropriate ai tool"]
c --> d["check quality, effort, and cost"]
d --> e["keep the improvement or change the approach"]
e --> f["apply what you learned"]
f --> a
if you lead a team, give people time to do this during work. provide access to suitable tools and clear rules about company data. invite the people who know the process to shape the experiment. telling everyone to embrace ai while giving them no time, support, or permission to learn is a poor way to lead a change.
for your own growth, choose something you actually want to accomplish. build a small application. practice explaining a difficult subject. work on a drawing. get feedback on a chapter. a project you care about gives you a reason to keep going after the novelty wears off.
you don't need to follow every release to make progress. learn how to define a problem, provide useful context, verify an output, and recognize when a tool is getting in your way. that experience remains useful when the product names change.
some people have already spent years developing it. if you haven't started, you may already be behind someone competing for the same work or pursuing the same opportunity. waiting another year will not give you back this one.
i believe the future is ai. i want to help build it, understand it, and use it to do work i'm proud of.
pick something that matters to you. start this week.