This is such a good gut-check — especially the point that a lot of “real” data work is days of hunting tables/joins and cleaning messy, undocumented data, then a comparatively small slice of modeling.
I also appreciated the boundary you named around AI: using it to scale what we couldn’t scale before, but not outsourcing judgment/ethics.
Curious: for aspiring data folks who *do* have the technical chops, what’s one simple exercise you’ve seen that reliably demonstrates stakeholder communication (beyond “tell me about a time” answers) in an interview?
Being able to clearly connect a problem to a solution that leads to a real business outcome or decision. You could call it STAR or CAR response, but it's a standard for a reason.
You mentioned that "AI is powerful, but overhyped", I'm not sure if it's overhyped, but I will say that there's a bit of a dichotomy in the message given by the top players of the field versus what the average company in a given industry will adapt.
For example, this week alone top talent at Anthropic and OpenAI said that they basically don't code anymore due to their products. This is obviously a marketing stint to raise hype that permeates to your regular Joe worker that's afraid about pursuing a career in coding.
But in reality I don't believe most small and midsize companies are going to take the measures to replace their workforce with Claude agents
I think a lot of what will hold AI back is cost and data availability. Not every company can afford to implement many of these solutions and won’t see the ROI. They also don’t have enough data to do much that would be useful. They might use some small out-of-the-box solutions, but for a lot of companies, there won’t be that much change, at least not for now.
This is such a good gut-check — especially the point that a lot of “real” data work is days of hunting tables/joins and cleaning messy, undocumented data, then a comparatively small slice of modeling.
I also appreciated the boundary you named around AI: using it to scale what we couldn’t scale before, but not outsourcing judgment/ethics.
Curious: for aspiring data folks who *do* have the technical chops, what’s one simple exercise you’ve seen that reliably demonstrates stakeholder communication (beyond “tell me about a time” answers) in an interview?
Being able to clearly connect a problem to a solution that leads to a real business outcome or decision. You could call it STAR or CAR response, but it's a standard for a reason.
You mentioned that "AI is powerful, but overhyped", I'm not sure if it's overhyped, but I will say that there's a bit of a dichotomy in the message given by the top players of the field versus what the average company in a given industry will adapt.
For example, this week alone top talent at Anthropic and OpenAI said that they basically don't code anymore due to their products. This is obviously a marketing stint to raise hype that permeates to your regular Joe worker that's afraid about pursuing a career in coding.
But in reality I don't believe most small and midsize companies are going to take the measures to replace their workforce with Claude agents
I think a lot of what will hold AI back is cost and data availability. Not every company can afford to implement many of these solutions and won’t see the ROI. They also don’t have enough data to do much that would be useful. They might use some small out-of-the-box solutions, but for a lot of companies, there won’t be that much change, at least not for now.