How long do you think ML will stay this hot? Recent real and promising ML application : 1. computer-vision (facial recognition etc) 2. NLP (machine translation etc) 3. Recommender system (ads etc) The application of ML in applications such as Alpha Go or even autonomous driving seems to be unrealistic in our real world. Waymo “s CEO just disclosed that he believe we need at least 10 years before we can achieve truly L5 automation in vehicle driving. The OpenAI”s Dota game seems to be a joke when they showcased it this summer at TI2018 With so much money going into the field, and so little breakthrough, do you think AI will still continue this hot? How long before this bubble will burst? PS: I got a second author publication in ICML 2017 and third author in EMNLP 2016
First author or it didn't happen.
At least 10 years to full L5 automation = AI and ML only gets hotter for the next 10 years. Now the majority of unsolved problems to reach full L5 are ML problems, mainly in scene prediction, handling of rare events, etc.
For me, ML was never hot. There is a very little amount of practical appliances for it, compared to attention it gets from media. People think they can make machines learn as humans do, but actually they're still using them for statistical calculations. It is not even close to how humans learn, it is even not in that direction.
What’s the difference? Automated pattern discovery and statistical inference is learning as far as most practical tasks are involved. Practical applications - looking at the tech space, what are the areas of advancements? Silicon die shrinking is one. What else? Look at anything from CES to advertising to new iPhones to platforms like Uber and, yes, Booking.com, ML in one form or another is core to that. In other words, you can count on one hand the number of new features being introduced that have nothing to do with ML. The op’s question could well instead have been “how long will tech stay this hot”.
We as humans "very likely" do pattern discovery and statistical inference for learning and decision making, we just do it in a much more sample efficient way, and we do so great in quickly detecting and memorizing rare events and anomalies. These are the areas that ML algorithms need improvements on, and that my friend is the reason that ML/AI research will remain hot (and crucial) for the foreseeable future.
ML for malware/anomaly detection is getting increasingly more popular too.
I agree. AI and ML has been talked about for many years and still only application of it is for predictive analytics, not prescriptive. Corporations will soon realize its a hyped up garbage.
AI has been this hot since the 60s, at least.
AI is the next natural step of computer programming. Think about it, programming languages are now great tools to enable human programmers "code" their intelligence. What is the next step?
ML is still primitive and a lot of research need to be done to achieve real AI.
Tech Industry
Yesterday
1496
Sell every stock you own and hold cash
Tech Industry
Yesterday
157
What is more difficult?
Software Engineering Career
Yesterday
540
How can I help my wife transition into tech?
2024 Presidential Election
Yesterday
785
Trump is accused of r*ping a 13 year old
2024 Presidential Election
Yesterday
392
Democrats will get trounced
AI is the way to replace software engineering. How long will software engineering stay this hot? Probably forever.