AI Trained on Old Scientific Papers Makes Discoveries Humans Missed

This is really cool!

4 Likes

Sounds like the beginning of the end of humans discovering anything, to me.

We were already on our way to that. There is still a lot to learn, but all the low-hanging fruit is long gone.

2 Likes

Doesn’t our definition of low hanging change with time and context and perspective?

1 Like

Certainly, but I think we have mostly exhausted physical discovery: fire, wheel, elements, fundamental particles, etc. are all water under the bridge. Biology should still offer some surprises, and I’d say epigenetics counts as a surprise from the perspective of 20-30 years ago. Even the number of new discoveries in biology will have a limit.

I’m not saying there isn’t a not of work to be dome filling in the gaps in existing knowledge, but new areas of knowledge are hard getting harder and harder to find.

I would be absolutely delighted to be proven wrong about this. :slight_smile:

1 Like

I’m not sure how any of that was “low hanging” fruit.

As you say, it would be hard to define what that actually means, so it’s not worth arguing about. In a universe with a finite number of things to discover, we will reach a point where discovery is no longer easy. Particle physics appears to have reached this point, requiring massive particle accelerators for experiments. Classical Frequentist statistical theory peaked in 1933 with the Neyman-Pearson Lemma.

I don’t think we are close to discovering everything there is to know in the biological sciences yet, but I do think that time will come.

1 Like

Really? How so? I have a hard time imagining this.

When you get to the point that you know everything that there is to know in the biological sciences, then evolution will happen and something will change so that you no longer know everything.

2 Likes

Or we might work out what other molecules would work instead of DNA/RNA/proteins to support life, possibly on other planets. Or we might find just life on other planets.

1 Like

This is something I’ve been thinking about for a while now. Up until about the Second World War, most of the “big” discoveries in science were made by single researchers working alone, with equipment that was readily accessible and that could easily fit into a single room. Nowadays, the “big” discoveries – and even some relatively modest ones – tend to be made by large teams of researchers and international collaborations, very often working with equipment costing billions of dollars down mines, up in space, or many miles in diameter, and involving petabytes of data gathered and analysed over extended periods of time.

1 Like

My assumption here is there are a finite number of things to be known, and some of those will remain beyond practical discovery. Given time we may discover all that is knowable, or reach some physical on ontaining observations on things we do not already know.

Yes, that would be at least one new thing to discover, but given what we know about evolution already, it is unlikely to be an entirely novel discovery.

That would be new too, but are there going to be infinite forms of life, or will we see similar patterns repeating themselves? IMO, the later is far more likely.

I’ve read speculation that some fundamental particles (physics) will never be directly measurable because if would require a particle accelerator the size of Saturn’s orbit.

Now once we start combining different things we know in new ways, that’s leads to a potential combinatoric explosion of new discoveries, but it’s still a finite number. We make also develop the theory to tell us what to expect from these new combinations (not sure how that would work, just speculating).

it is cool but only because it corrects human incompetence. Its still all about memory.
All it can do is stop where a logical/math conclusion should of been drawn on options after some conclusion was made BUT was carelessly missed.
Its just correcting the math. In fact it hints that science is all about conclusionsand then simple more conclusions from the original one.
Another thing IT would do is make impossible claims from where conclusions were wrongly made.

Humans have figured out very little about most subjects. the easist ones went first like obvious physics,/Newton/Einstein, however as it gets more complicated a curve of difficulty is made.
Biology is the most complicated and thats why evolutionism was so dumb to jump to conclusions without accurate scientific investigation.I think very little has been figured out yet.

1 Like

I agree – and I think that something like this is greatly needed with all of the highly specialized technical papers that we now aim to churn out.

Yeah the past 150 years of conclusions about evolution have had little scientific investigation beyond hundreds of thousands of papers.

2 Likes

Thats just the point. The presumptions behind biology conclusions was not well done. Any papers just repeated the presumptions and worked from there.
niology is complicated and must have well done science. Evolutionism was a guess, agreed to right away, based on lines of reasoning on very raw data. evolutionism has no claim to being a biology scientific subject as I see it. its a untested hypothesis.

The accomplishments of science for the betterment of mankind are numerous - life expectancy, food production, clean water, sanitation. Compare life during biblical times with today.

2 Likes

Key phrase ‘as you see it.’

1 Like

This may just be the tip of the iceberg!

https://www.nature.com/articles/d41586-019-02142-1

1 Like