Pre covid, I remember seeing "ML enthusiast" on many LinkedIn profiles, especially every university grad student, but I don't see them anymore. Are we finally back to earth now ?
Yup. Dead in the water.
I think so. When everyone realized that 90% of data science is cleaning up data.
If you're talking about ML hype as in the next sexy startup which claims to "solve your business problem using ML at the click of a button", then yeah people are waking up to the fact that it isn't that easy. But there are strictly more products in Google that are running ML today than say 2 years ago, so while the hype may be decreasing, the actual usefulness is increasing.
And the capital keeps flowing to a few selected big techs. Startups realize they can't compete with Google and Microsoft on the state of the art NLP. Other trivial problems can be done by a SWE and don't require a real ML scientist.
This is absolutely correct, and it's a tragedy because it's difficult for non-FAANGM companies to compete in this space because they simply don't have the data to begin with
Big Data, ML/AI. I wonder what's coming next
The dust is settling down, the serious ML/AI work with great business impact is still highly rewarded and experts are sought after.
Examples? The concept is useful but when someone like Netflix use DNN to reccomend me 100 action films just because I watched Die Hard is a good example of why it can suck for some businesses. I don't sign up to Netflix to see the same crap, I sign up to see the variety.
Do you think that's something new with NNs? Netflix has recommendation systems since the beginning. All the large scale firms get 100s of millions of dollars in additional revenue for a 1%/2% improvement in their ads/recommendations/search systems - they will continue to hire top tier ML scientists to push the envelop - while small startups will get basically nothing on top of some standard implementation.
Depends. I think the industry has delineated the roles nicely now, compared to 4-6 years ago. ML Scientist: Need a PhD, not dead at all ML Engineer: Need to understand some ML, Stats but mainly SWE, not dead Data Engineer: Need to understand data Engineering concepts, dist systems, Spark/Hadoop, Scala perhaps, Absolutely NOT DEAD, growing in fact Data Science: Dead dead dead
I was in a rare spot of being a Sr HRBP for 2 companies with ML exploded. It was a fascinating thing to see - I feel really fortunate I got to see a market shift in action. I mean just watching the data come in, the job profiles update, the trends and analysis was so effing cool to see. Coming from an internal POV - the "hype" was definitely chaotic. In one of my FAANG+ companies, the head of Tech was a man on a mission to scoop up as much ML as possible and he and I drove each other crazy because of our views. The problem was we were hiring before we knew what do with them. So we'd have these amazingly brilliant ML engineers and be like "well...you are here....go forth and do the M to the Ls! and a month later: "Did you build that really cool think we can show to the CEO yet?" "What thing?" "Your Machiney Learney thing!" When I left, they were still very actively pursuing ML, but Im seeing a lot of people now be more thoughtful with the approach. It's a highly valued market skill for now with a clear case for demand and a limited supply, but definitely not crazy like it was. Now the cool hip LinkedIn profile to be is "Career Whisperer" and "Audience Engager"
The problem is companies just threw money at it after some McKinsey idiot showed them a graph saying ML was the next big thing. They hire people and have no strategy. Complete hype.
The companies that know how to use ML (Google / Amazon) are continuing to hire and build with it. Everyone else seems to be building simplistic crap that they would be better off copying from github.
I do agree that ML is hyped a lot but some of it is warranted. In FB and google ML based recommender systems have outclassed other algorithms and as a result has contributed to higher ad revenues for both. In Computer vision, ML algorithms are state of the art and same goes for NLP as well. ML models are also state of the art for spam detection. As you can see ML has revolutionized several fields and will continue to do so.
Tech Industry
2d
27187
How did this happen? (Meta Stock)
Tech Industry
Yesterday
2530
Tech companies to avoid as a white guy?
Tech Industry
2d
1028
Is 2.5M enough to retire in Bay Area?
Tech Industry
Yesterday
2015
1 vs 5 Million - no lifestyle change
Tech Industry
Yesterday
1758
Lack of diversity in engineering division at X
Hahaha. Typical Hype cycle.
Yep, it’ll be back again in 10 ish years or so