Statistics Are About The Past. Look To The Future.

Sunday, 7.54pm

Sheffield, UK.

I abhor averages. I like the individual case. A man may have six meals one day and none the next, making an average of three meals per day, but that is not a good way to live. – Louis D. Brandeis

I spent nearly three hours yesterday working and reworking a handful of sentences, trying to articulate what I had done between 2013 and 2017.

I wrote sentences. I wrote some more. I looked at them, shuffled them, reordered them. I put them aside. Later, I wrote new sentences on the same topic. Now I need to read, reorder and eventually type them into the computer.

What’s the point?

In five minutes between starting this post and writing the first paragraph ChatGPT wrote me a 1,759 word briefing note on logistics decarbonisation. It’s easy to read, contains what I need to know, and can be used virtually unaltered.

How can you be sure it’s correct? There are facts in there that I haven’t checked. But the weight of probability is on the side of the machines – literally – because it’s a statistical machine and the words are the most likely ones that would come up in that kind of writing. The last time I corrected the output from a generative AI, I introduced mistakes. It felt like a milestone, where I, the human, was the most fallible part of the system.

So where is the role for humanity in this? Do we just sit back and let the tide of AI generated material wash over us? How should we respond?

I think we do two things.

First, we recognise that knowledge is interesting and wrestling with data and information is part of the process of acquiring knowledge. If tools make it easy to do parts of the work then that simply means we can extend the edges of what we can learn about. When books first came out people complained that there were too many books being published to read in a lifetime. The Internet exploded our access to content. Now generative AI can spew out unlimited amounts of material.

In responds, many of us will pull down the barriers and restrict our reading to trusted material. How many news sites do you go to now? Our capacity for information processing has not increased with the technology for information creation. So we have to be selective – use our attention intentionally to gain the knowledge we are interested in and need.

Which takes us to the second point.

Statistics are good for showing what happens on average but not what happens to you in your particular situation. You can focus on your niche and figure out what you need to do to add value in a specific area. That area will be big enough to be interesting and valuable and too small to be able to be tackled with statistical methods. It will require human involvement, armed with tools like negotiation skills and constructive approaches – where you and others co-create the future.

In other words, where you look forward.

Statistics is about the past.

Human work is about what comes next.

Cheers,

Karthik Suresh