Serious question. SAS has long touted itself as being a leader in AI. Should SAS have been the ones to bring ChatGPT capabilities to the market, or are we just not built that way or have the innovative capabilities to think bigger than we are.
30 replies (most recent on top)
As someone working in the space. Let me offer my two cents.
This explosion of "AI", especially for text, really only begun to start (I would argue) with the release of the "Attention is all you need" paper in 2017, which introduced the famous Transformer architecture. At the time, Google was still at the top in terms of industry leading research and development. Since then, there was steady improvement and constant new streams of models, techniques, ecosystems, etc. ML/AI was viewed as more of a niche topic and field, especially compared to nowadays and the public buzz that it has generated.
Obviously, when OpenAI released ChatGPT, it was like the floodgates for this new AI craze opened, and everyone wanted a slice of the AI pie. For once, it seemed that an AI product was able to "break" that tech bubble and enter the general psyche of the public, and as OpenAI showed, there was a marked interest in it, and importantly, money to be had.
Also notice how major industry players, like Google, failed to capitalize early on with their discovery of the transformer, and let OpenAI take the initiative (with MSFT riding their coattails). I bet they're kicking themselves trying to play catchup.
Now as for SAS. To be completely blunt, in my opinion, we are not built whatsoever to capitalize on this AI craze. Both in computational resources (cough cough, GPUs, which are insanely expensive, thank you Nvidia), human resources (there is a limited amount of people skilled in this area that are able to develop, innovate, and test. Even more so when trying to take those research ideas and put them in a product), and management capabilities. I want to stress the management capabilities here.
I actually think, believe it or not, that with the proper resourcing, there are skilled enough individuals here at the company with a vision that can produce a worthwhile SAS AI product -- especially when we have the ability to leverage the sheer amount of SAS that has been written by our testers and SAS devs. However, that will not happen, and this opportunity will perish from a lack of proper utilization and planning by upper management. Throughout this nascent phase of trying to spin up AI products to enter the market with, there have been disgusting examples of managerial incompetence or petty squabbling, all while we lag further and further behind the market and our competitors, who, day by day, make further strides in releasing AI products, innovating with new techniques, enhancing AI security principles, etc.
I continue to have hope in my coworkers and the immediate management I work with, but yeah, it is disheartening.
I think what @dblz+1sXiXn29 is saying is that they don’t believe this was actually said and this is just another case of anonymous people saying crazy sh-t.
But it should be easily verifiable if it was a company wide meeting…
@9kkh+1sXiXn29 Would love to see transcript of that meeting to see exactly what was said.
Care to say what company wide meeting?
"I can’t tell whether they were trying to be politically correct or were truly clueless."
They were gaslighting.
Answer: The one man that was capable of advanced thinking was busy running the company.
@9kkh+1sXiXn29 “too ethical”
I’m sorry that person felt the need to lie. It’s not even a competent lie: no one would believe it.
They could have just said: we have dozens of employees working on AI, while other companies have hundreds. It’s a sufficient explanation.
I can’t tell whether they were trying to be politically correct or were truly clueless.
lmao. "too ethical". mgmt has no clue it has no clue. remember pfeffer and others lauding sas' hr "montessori" policies? it really seemed believable and sensible at the time. at some point, you need some actual top talent not living in an insulated bubble, drinking the kool aid.
"too ethical" to innovate but not too ethical to layoff.
Interestingly, this very subject came up in a company wide meeting. The person in charge of that division assured everyone that SAS could have create ChatGPT easily, but that as a company we were "too ethical" to it. AFAIK, that person is still in charge of that division and SAS is still "too ethical" to innovate.
@5fdd+1sXiXn29 Not sure why I'm even bothering to argue with someone who claims to know the certification and technology mastery status of every single person in SAS R&D.
And then sticks to those g-ns despite being told he is wrong by someone who knows specific people with such certs and skills.
Some simple google searches that validate what I've already told you without me having to name people I specifically know and dox myself.
The google results are far from perfect but you can do a tiny sliver of sifting to see how wrong you are.
site:linkedin.com AND "SAS Institute" AND "Certified Kubernetes"
site:linkedin.com AND "SAS Institute" AND "Microsoft Certified Azure"
I know that probably goes beyond the 2 minutes you spent in cloud academy.
And for what its worth there was nothing wrong with the cloud academy material. There was a sh-t ton of it if you put in the effort. Some of it good. Some of it not.
The effort part is on you not some nameless director.
If you chose to only do the required material then yeah that isn't going to get you anywhere... Take some initiative maybe.
@5fdd+1sXiXn29 I have every idea what I’m talking about with regards to Azure certifications, Terraform, Ansible mastery, and the CKA program.
Just because you didn’t do any of those things doesn’t mean nobody else did. I know for a flat out fact that people have. I’m not throwing out names but I have some.
You are just wildly generalising based on no actual knowledge.
Yah, no. The Cloud Academy training the former director of DevOps R&D threw together didn't feature more than an introduction to any of those technologies, and wasn't specific or focused enough for anyone to then go on to get any certification. The Azure certification alone takes hundreds of hours. You have no idea what you're talking about.
@4exz+1sXiXn29
“How many employees became "Microsoft Certified" as a result of all that training? The answer is "none" “
Wrong
“ How many mastered Terraform? None.”
“How many became proficient at Ansible, or earned their CKA? None again.”
Wrong
“ And why is that? Because those skills are portable”
Wrong since all the things you stated leading into it were wrong.
I really can’t even keep reading past that. Those things are not wrong as in we have different viewpoints. They are demonstrably and quantitatively wrong.
Just keep making sh-t up. Eventually you’ll be right.
Wow. How toxic.
SAS doesn't invest in its employees. Case in point: the Big Plan in DevOps R&D was to hire in a bunch of really smart "cloud engineers" because none of the "akshully stoopid" employees doing the work of the company could be spared to learn anything new, unless they were functionally useless or one of the director's drinking buddies. Those positions never materialized, and the people with the Big Plan have all, with the rare exception here or there, been invited to pursue opportunities elsewhere, so vital to the success of the company were they.
You may be asking yourself why with all the Cloud Academy "training" that was assigned to everyone in the division, none of it was specific to the cloud technologies that SAS was publicly adopting. How many employees became "Microsoft Certified" as a result of all that training? The answer is "none". How many mastered Terraform? None. How many became proficient at Ansible, or earned their CKA? None again.
And why is that? Because those skills are portable. Those employees would have had a ticket to leave SAS for a better job somewhere else. So of course the "training" was a bunch of useless home-grown junk, courses for solutions or technologies SAS doesn't actually use, or training so generic as to be practically useless.
On the other hand, it created the checkbox that was used to punish the principal or distinguished testers that were laid off in August of last year, and who "hadn't kept their skills up-to-date despite ample opportunity to do so". And after the layoff, BH had cake. So I guess it wasn't completely useless.
And then there are the toxic apologists that think those people deserved their fate, because they have to. They have to believe there's some reason they have been spared... some intrinsic knowledge or skill they have that makes them valuable. Maybe it's their superior work ethic. They're only escorting the corpses to the hangman, you see, they're not pulling the lever themselves.
But the joke's on them. When the hangman has gone through all the other "undesirables", he'll come for them too, then dust himself off, wash his hands, and go have cake.
@4qmo+1sXiXn29 “I can hear the Platters, or Freddy Mercury singing it now: "Oh yes, I'm the great, Defender..."
Hear it all you like. It isn’t somehow “defending” to say it is naive to think tech companies don’t get out of date and n need of modernisation.
If our average employee fits the average poster here then that is probably just as reasonable a reason for SAS being in trouble. Shallow thinking skills at best.
"If you start from the beginning in the 70s then SAS predates the work Bill Gates did in his garage. They could have been Microsoft, not just CharGPT."
Jim Barr started SAS in the 1960s. But, wow, point well taken!
"You should start your own company."
I can hear the Platters, or Freddy Mercury singing it now: "Oh yes, I'm the great, Defender..."
If you start from the beginning in the 70s then SAS predates the work Bill Gates did in his garage. They could have been Microsoft, not just CharGPT
if you only sell on-premise software tools, you're not in position to have a "big data" business and build a "data moat" and then be able to have models trained on the large data and build the "big data" and "AI" competencies in house. selling tools (for customers to run on customer's on-site hardware) doesn't build those capabilities or potential. msft, perhaps the best or only example of an on premise vendor successfully shifting to cloud, made transition moves such as shifting sharepoint and exchange servers to azure while keeping end user productivity UIs the same familiar tools on clients, protecting its user base while building cloud capabilities behind the scenes, and solving a backend IT need. sas ... made ... other decisions - well documented in these threads.
ChatGPT and other LLMs are the result of a research breakthrough (https://proceedings.neurips.cc/paper_files/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf). Most tech companies did not realize its implications.
So most tech companies got caught by surprise and are now frantically “infusing” their products with AI. Even Apple announced yesterday that they’ve licensed OpenAI’s technology.
SAS, or any company looking to sell or IPO, needs AI to make their products look modern and attract investors. I don’t fault SAS for being behind the curve on AI; most companies were.
I do fault SAS for being behind the curve on open source and the cloud. After they’ve sprinkled AI pixie dust on everything, these architectural problems will remain.
@1bia+1sXiXn29 “ Tech companies should never be behind the curve and require modernizing. What a train wreck.”
That is such an unrealistic and naive comment. Of course technology companies get behind the curve and have to modernise. Not every company is Google. And even Google has lots of pockets of old sh-t running.
You should go start your own company and get back to me in 10 years.
Others are doing exactly that, yes. But these other companies aren't companies that sat on their laurels for the past 10-15 years and then suddenly scrambled to wake up and try to be competitive.
It's just too late now. Why would any self respecting technology company go around talking about how it's modernizing??? Tech companies should never be behind the curve and require modernizing. What a train wreck.
You don't have to be a producer of LLM's "infuse" AI into products. You can partner with the providers of LLMs. ServiceNow, Salesforce, Oracle, Adobe, and many others are doing exactly that.
So there's no reason why SAS couldn't "infuse" AI into it's products as it claims it's going to do. There are a few obvious use cases, like code generation from natural language, predictive model building from natural language, report generation from natural language, etc.
However, here's the problem. Nobody wants SAS's products anymore. So even if SAS does a great job at "infusing" AI into it's products, they will still be cr-ppy products, built on a flawed architecture and an antiquated programming language, that does stuff that open source does better and for much less cost.
The directive is to "infuse" AI (read: force fit it) into everything everywhere at all times. Let's not dwell on the fact that we don't even offer an AI product. The company is out there saying AI and analytics in all the marketing material and PR and videos but we don't actually have anything to back it up. Ridiculous. Blame the big guy and the circle of yes people around him. Nobody wants to tell him he's not wearing any clothes.
" I remember laughing when SAS announced a 1 billion dollar investment in AI. "
The $1B went to the Art Department for the "Art Initiative" project.
Years ago, PR worked to get an article naming JG the "Godfather of AI" as if it all started with him. It was such a joke, and an insult to anyone who studied Computer Science. Marketing spun it and it was a flop outside of the SAS Vortex. They reference the article on his Wiki page, but no one else calls him that.
https://www.cnbc.com/2019/11/04/godfather-of-ai-predicts-the-future-fate-of-the-us-workforce.html
https://en.wikipedia.org/wiki/James_Goodnight
I thought SAS was the analytics king? Is that no longer relevant so now its gotta be the AI king? It feels like just spewing out some buzz words to stay relevant. Hackathon anyone?!
+1 to each of these responses.
I was close to the core group that would/could have been responsible for this innovation. I remember laughing when SAS announced a 1 billion dollar investment in AI. I joked with colleagues that half went into building A and half would go into a future building I. :)
All kidding aside, this group had 10-15 folks. They were asked to cover Image and Video Analytics, Text Analytics, NLP (understanding and generation) and Chatbots. Pick any one of these and other companies likely employ 100's of engineers.
Then you read that the reported cost to generate some of the new GPT models is in the millions of dollars for one training run.
At the time, it was all we could do just to stay abreast of all the innovation coming out of the big tech companies while doing our day jobs.
Could SAS have been OpenAI? Yes, but we would have had to divert most everyone to the effort as well as hiring some massively smart people in the field. So for all practical purposes, no.
@n-g+1sXiXn29 has it right. SAS needs the "AI" buzzword, not only for publicity by Gartner, but also to attract potential buyers or IPO investors. So SAS does a few small AI projects, enough to be mentioned.
Serious AI takes serious money. Microsoft invested $13B in OpenAI. Google and Meta are playing catch-up, and they have enough money to play that game. SAS does not.
SAS should have stuck to its core competence, data analysis and business intelligence. We let Tableau, PowerBI, and others innovate and take market share. But we had the expertise to compete on our home turf.
SAS was never relevant in the "AI" space. SAS releases statistical modeling software, so I can see how someone would imagine that it's just a short leap to AI. But AI was (and still is) just a buzzword someone throws out occasionally to appease Gartner and make sure SAS gets a mention somewhere in the "Magic Quadrant". Actual AI research requires a massive investment in data aggregation and compute resources. R&D has been scavenging hardware from the discard pile for a decade now. You can't just want to "do AI", you have to actually put your money where your mouth is.