Originally published at: This AI project distills research papers into a single sentence | Boing Boing
…
Interested in the results, but concerned about the prospect of more vaccine-deniers citing even more studies without actually reading them.
So I put in some of my articles, and wow is it bad. None of the resulting TLDRs was even a full sentence, let alone a useful description. These are the results:
“We show that a collision-detecting neuron can detect the spatial coherence of a simulated impending object, thereby carrying out a computation akin to”
and
“We show that the membrane impedance of a collision detection neuron in the grasshopper Schistocerca amer- icana is strongly influenced”
and this, which is half of a random sentence from the intro
“A collision-detection neuron in the grasshopper exhibits a large M current, generated by noninactivating KCNQ/Kv”
and the clear winner is this wonderful brilliance
“Neuromuscular Facilitation in Crustacean Spiders and Shallow Shores ."
Isn’t that what an abstract is? A concise summary, possibly even translated into an additional language for the benefit of Quebeckers?
Now I want to read both your article and these alternate reality versions!
What’s a crustacean spider? Is that AI babble or an actual thing?
I’d like to read the aleternate reality article about the “Crustacean Spiders” for sure. What would that even look like?
Definitely not real. Spiders are arachnids which is completely separate from crustaceans.
The actual article was comparing a species of spider crab and shore crab. The AI turned spider crab into “crustacean spider” and shore crab into “shallow shores”
I wondered if that’s what happened. Too bad. The idea of a crustacean spider is both horrifying and super cool.
This project says
Currently, our model is only trained on English-language papers in the Computer Science domain.
Naturally, my thoughts turned to the problem of counterfeit conciousness
Thus,
Input Text
Adaptive control of the research object is the uncertainty of the system to a certain extent, so-called “uncertainty” here refers to describe the mathematical model of controlled object and its environment is not entirely sure, including some unknown factors and random factors. Adaptive control and regular feedback control and optimal control, is also a kind of control method based on mathematical model, the difference is adaptive control is based on a priori knowledge about the model and the disturbance is less, the need to constantly in the process of the operation of the system to extract the information on the model, the model gradually improve. This topic in the existing research for adaptive control mode on the basis of the specific application scenario for the related algorithm used in motor control, motor control is to point to, for the start of the motor, speed control, operation, slow down and stop. According to the use of different motor types and have different requirements and purpose. For motor, motor control, to achieve quick start motor, fast response, high efficiency, high output torque and overload.
Generated TLDR:
“Adaptive control of the research object is the uncertainty of the system to a certain extent, so-called “uncertainty” here refers to”
Damn.
And who would use such a thing? Scientists too dumb to do science?
Apparently, they can be condensed down in to a single word.
Scientist who wish to cast a wider net. Consider
In one day, 195 papers were announced. Do you want to read 195 chatty abstracts per day to figure out if just one or two will be possibly be relevant to your research?
The abstracts aren’t “chatty”. In my experience they’re just brief enough to let me know whether it’s potentially useful or not. And because 195 abstracts is about as much text as a novella, that would indeed be a burden on me, if I didn’t know a smarter way of doing research than reading 195 random abstracts per day.
Like I said, scientists too dumb to do science.
I prefer white meat.
Watch through the Q&A.
This topic was automatically closed after 5 days. New replies are no longer allowed.