Experienced multiple times working on applied AI projects. Real-world datasets are messy and labeling them is an expensive task.
Enterprises are also not sure how to measure the RoI of their AI projects, especially since the accuracy is the mid-80s at best and someone needs to take ownership to teach the machine and improve the accuracy continuously.
Great - wonder how do you guys hold sway amidst tough competition from Amazon and Google for text analysis solutions?
It's a tough world out there for startups in this space.
If someone can assure me that my data will be secure while using these cloud based solutions, I will be happy to pay a premium for accessing such services.
Unfortunately, both Google and Amazon provide no such guarantee which makes it difficult for a financial organization like mine to use it and therefore, I feel if you guys can provide an on-premise solution in the form of an excel plugin, that could be killer!
Hehe! You will be surprised how often this happens in best of the companies.
A large fintech startup I work with had 5 GPU servers lying around idle before the CTO realized that they only need them while training the Machine Learning models and not during inference stage.
Been there. Remember getting a call from a data center I'd never heard of asking to speak to the old sysadmin who'd quit 18 month ago. They asked something about the rack of servers we had in their data center, I replied "what rack? what data center?" Turns out that 2-3 years previous the old sysadmin had set up a dozen servers in their data center for some project (that I'd never heard of) that very shortly afterwards first got pushed forwards and then canceled. Everybody involved in that project promptly moved on to other things and no remembered to cancel the deal with the data center. So for two years we'd been paying quite a lot of money to keep a completely idle rack of pretty expensive servers and apparently no one noticed.
Very classic one is x.ai or Amy or whatever they call it now. They overhyped their capability of building a smart assistant for reading and replying to emails and ended up being an average service that sort of works.
Earlier, they kept telling me that I have jumped their queue and will soon get access. Now they ask me to upgrade the plan to keep using it. Complete BS!
Average? I've used it at least a dozen times in the last month, and it's worked perfectly every time. It may not be the fastest, but it's certainly the best... YMMV, of course.
It's good but considering auto-reply feature of Gmail and integration with other apps like Calendar and Hangouts, I wouldn't pay for any such solution.
Of course, assuming the organization use Gsuite else, maybe x.ai fits the bill.
Sendy's templating engine is a mess, best that worked for me is to design the template in Mail chimp, export it as HTML (Mail chimp provide this as free of cost) and then, use sendy for sending mails.
Since so long, I have been waiting for Indian universities especially IITs to invest and publish in building such corpora.
Being a founder of AI/ML startup, I am surprised at the appalling lack of datasets available to work on Indian problems. Contrast this with Chinese universities where they have built some world class datasets to build NLP solutions in Mandarin.
Our sentiment analysis works in 8 different languages but none of it is in Indian languages despite we being in India!
CC-BY-NC amounts to saying: you can play around with this in demos and academic projects that no lawyer would ever go after anyway, but you can't use it for real.
"Non-profit" is a whole different kettle of fish than "non-commercial". It doesn't mean you're not selling anything. It doesn't mean you're morally good. It just means you registered for a particular business status with particular restrictions. And it's not what CC is talking about.
A key part of the Open Source Definition is that you do not lose your permission to use and copy the code based on what you do with it. A project that you aren't allowed to use anymore if you start making money from it is not Open Source.
"Non-commercial" code restrictions are more like the thing where you're allowed to look at the Windows source code, if you're an academic and you ask nicely and you won't ever do anything with it.
A lot of universities are open to give a separate commercial license when contacted. They charge for their efforts, which is fair. Whether public universities should charge given we already pay them from our taxes is a different issue.
source: I am also cofounder at an AI startup, we often buy licenses to use academic datasets for commercial usage.
We have trained our model on public photos which were popular/trending at some point of time. We fed these images to a deep CNN (Convolutional Neural Network) which started to recognise features that made photos popular. What we realise in the process that these features do change with time so we have added a temporal component to our training set to ensure our model is relevant.
It is currently optimised more for human photos and travel images so maybe in our next iteration, we can predict pets photos as well.
Once you create the test, you can open it up for applications. You can also search based on the demographic you want (age range, geographic, etc.) and invite users yourself.
If people apply, you can pick from those. I invited about 60 testers and had ~180 others apply. I was able to pick from that 180 set - and the reward goes only to them.
Also, before you start the test, you buy 'Beta' credits. You can choose to use these however you want, for example if you buy 50 beta credits, you can give a $5 reward to 10 testers, or a $10 reward to 5 testers. You can buy more credits as you go if you want to offer rewards to more testers.
Enterprises are also not sure how to measure the RoI of their AI projects, especially since the accuracy is the mid-80s at best and someone needs to take ownership to teach the machine and improve the accuracy continuously.