Gus Bekdash

Director of Enterprise Architecture @NTT Always thinking about how to transition from where we are to where we want to be in #IT, #AI, #Technology & life.

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr Are you doing anything suspicious :)

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr No doubt some humans have more insight than AI.
But I am not convinced humans have more insight generically.
AI is evolving so fast, I think we need a model where we don't have to exceed AI at all in any area, other than being the boss.

[Gus Bekdash @Gus_Bekdash]

Some CIOs are starting to say this to me (between the lines). There is no real advantage to being a pioneer in AI. It's best to wait then jump when a specific approach is proven. @levie How do you weigh in?
#CXOtalk

[Gus Bekdash @Gus_Bekdash]

I think @levie 's idea for an "internal forward looking engineers" (FDE) is brilliant. It's not just for vendors or consultants.
#CXOtalk

[Gus Bekdash @Gus_Bekdash]

So true about the LLM progress. But The last few months seem to change NEW things a lot in terms of bringing LLM AI more context and ability to connect with workflows, not just raw power. Do you share that opinion?
#CXOtalk

[Gus Bekdash @Gus_Bekdash]

True for many things but Esp agentic.
Some declare amazing victories ... because their objectives were so modest. How do you determine the right level of ambition to avoid irrelevance by being too small or too large in AI and automation in general.
#CXOtalk

[Gus Bekdash @Gus_Bekdash]

Aaron, every exec I talk to in IT defines agentic AI differently. And the differences are HUGE. How do you think about it?
#CXOtalk

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr Very good question, in a good framework. But isn't what it's supposed to value be determined by us? I don't think it should be or can be random. TO me the questions are:
1. would we misfire (aim for X and get Y), 2. can it change itself, 3. wouldn't some humans want evil AI?

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr This is great Tom, introducing up to our own state, which regrettably, we don't know as well as we should. Thanks.

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr Looking great Tom! Have fun.

[Gus Bekdash @Gus_Bekdash]

Most of you are willing, unwilling but mostly sleeping passengers in AI Air. Most AI captains share my flying skills: know how to take off, but not navigate, land, or recover from a stall. Good luck.
#AI #AIstrategy #AIinvesting

[Gus Bekdash @Gus_Bekdash]

@cxotalk @sanguit @ArsalanAKhan @Scriffignano1 @CPetersen_CS Thank YOU @mkrigsman. It was a very good session, as always. And thanks @sanguit for very good ideas.

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr Tom, you're going for objective reality. But some will fight you because of this: history is loaded with examples where "the narrative" shapes reality just as the opposite is true, & AI will drive the narrative, and they want the CEO to shape the narrative their way.

[Gus Bekdash @Gus_Bekdash]

Your AI team is telling you: “Whatever we do, someone will call us biased.”

You are the CEO. What do you do?

As #AI affects more decisions, this WILL become your problem. 1 / 2.

#AI4Leaders #AIEthics @tomyoungjr

[Gus Bekdash @Gus_Bekdash]

In AI ethics there are no clean choices, only tradeoffs.

1️⃣ Match reality and you look unfair.
2️⃣ Match “fairness” and you distort reality.
3️⃣ Tradeoff, and everybody will hate you.

The real decision is which bias you own and why. 2/2

[Gus Bekdash @Gus_Bekdash]

@tomyoungjr Astute observation on where things are, @tomyoungjr

[Gus Bekdash @Gus_Bekdash]

@OpenAI's real strategy is no longer about their AI! They are using #AI to get to an AI compute business that they think is more sticky, as LLMs converge in capabilities. #cxotalk

[Gus Bekdash @Gus_Bekdash]

@Sanguit, you're a very smart man! Very good points. #cxotalk

[Gus Bekdash @Gus_Bekdash]

If we examine science history, we have not had REAL breakthroughs for a while. There are no easy problems left for our scientific method, which is geared to complicated, not complex, SYSTEMS problems. I believe #AI & #datascience will change that. #CXOTALK

[Gus Bekdash @Gus_Bekdash]

The rigid ROI approach limited #RPA (more than it's inherently limited). Applying that mentality to #AI is even worse! #cxotalk