Year: 2021

The progress CG artists are making with Epic’s MetaHuman Creator is astonishing. I’m seeing really exceptional things on Instagram: https://t.co/O5aFySYZfw

Building something easy for the vendor vs. building something effective for the customer.

@ashwinmegha The #NoOps concept is 10 years old. It didn’t happen. And we are not even remotely close to getting there. Why it failed and how to move forward:
https://t.co/gI7RBY2nVJ

Yoshua Bengio, Yann Lecun, and Geoffrey Hinton on the future of #AI and the challenges to get there: https://t.co/IUcc6vkqOd https://t.co/j1AwoWjiol

5/ Monitoring + #automation is the only way to move from “admiring the problem” to “addressing the problem” at the scale IT environments reach today. If this is hard to act on, imagine monitoring feeding #AI feeding automation. And yet, both opportunities are right there.

4/ For this to happen, #automation engines should integrate with all sort of monitoring/logging tech, which is apparently impossible to achieve. After 30y monitoring vendors and automation vendors are still so far apart that they can’t see the benefit of working together.

3/ The real “intelligent automation”, the big breakthrough would be if #AI could learn in an unsupervised way what has changed in the aforementioned ever-changing IT environment and adapt the existing #automation workflows accordingly. That would dramatically lower the TCO.

2/ We often talk about “intelligent automation” in terms of #AI coupled with #automation to execute whatever decision the algorithm has come up with. That’s improper terminology. It should be called “applied intelligence” or “operational AI” or something else.

1/ Rarely discussed, the TCO of #automation is largely impacted by the enormous effort necessary to maintain existing workflows relevant and functional in an ever changing IT environment. Just like #AI, automation is not a deploy & forget technology.

A sombre report about the state of applied data science in the enterprise. In the last 3y, I said repeatedly that there is a complete disconnect between skills hired in #AI R&Ds/BUs and skills needed to understand what large end-user orgs need.

https://t.co/JFrHNCFFHs