I'll be finishing my interventional radiology fellowship this year. I remember in 2016 when Geoffrey Hinton said, "We should stop training radiologists now," the radiology community was aghast and in-denial. My undergrad and masters were in computer science, and I felt, "yes, that's about right." If you were starting a diagnostic radiology residency, including intern year and fellowship, you'd just be finishing now.…
GPT-4
531–540 of 1001 posts
Re: GPT-4
#532He nonchalantly announced that GPT-4 does "mental math" and then just didn't elaborate on that. Is it just me or is that absolutely incredible? What does that tell us about the underpinnings of language and mathematics?
Re: GPT-4
#533---
In the year of twenty-eighteen, Alice and Bob, a married team, Their income combined reached new heights, As they worked hard day and night.
Their son Charlie was their joy, A little baby, a lovely boy, A household they maintained together, Yet lived apart, without a tether.
To calculate their tax, it's true, A standard deduction we must construe, For married folks who file jointly, Twenty-four thousand dollars, quite pointy.
Their income sum, seventy-eight thousand nine eighty-one, Minus the standard deduction, the math's begum With exemptions being zero, the next line we trace, A taxable income of fifty-four thousand nine eighty-one takes place.
Now to the tax table, a liability we seek, For married couples, the outlook's not bleak, In range of thirty-six thousand nine to eighty-nine thousand one fifty, The formula's set, no longer shifty.
Five thousand five hundred thirty-five, it starts, Plus twenty-eight percent of the excess imparts, Eighteen thousand eighty-one, the difference we find, Multiplied by point two eight, the tax liability's designed.
Ten thousand five hundred ninety-seven dollars and sixty-eight cents, A tax liability for Alice and Bob, a sum quite dense, In this world of numbers, a story unfolds, Their financial journey, in a poem it's told.
Re: GPT-4
#534This technology has been a true blessing to me. I have always wished to have a personal PhD in a particular subject whom I could ask endless questions until I grasped the topic. Thanks to recent advancements, I feel like I have my very own personal PhDs in multiple subjects, whom I can bombard with questions all day long. Although I acknowledge that the technology may occasionally produce inaccurate information, the…
But it often produces wrong information. If you don't know the subject (since you are learning), how do you distinguish between correct information and incorrect but very plausible-sounding information?
Some random redditor ended up figuring it out. Then every physician from that point forward agreed with the diagnosis.
Licensed based medicine :(
Re: GPT-4
#535Re: GPT-4
#536Test taking will change. In the future I could see the student engaging in a conversation with an AI and the AI producing an evaluation. This conversation may be focused on a single subject, or more likely range over many fields and ideas. And may stretch out over months. Eventually teaching and scoring could also be integrated as the AI becomes a life-long tutor. Even in a future where human testing/learning is no l…
In essence, this tool will eventually allow us to scale things like private tutors and make educators more productive and effective.
We already have really convincing text-to-speech and really good speech recognition. It won't be long before we pair this with robotics and have lifelike tutors for people that want to learn. Kids of the near future are going to be so advanced at scale compared to any previous generation. A curious mind needed to have smart adults around them willing to get them resources and time. Soon anyone with curiosity will have access.
Re: GPT-4
#537Can't seem to find basic information like how many parameters were used or how big the training set was. Results are very impressive but would like to know what they are coming from!
Re: GPT-4
#538Live demo happening now! https://www.youtube.com/live/outcGtbnMuQ
Essentially, it's like a (text only) replicant
https://en.wikipedia.org/wiki/Replicant
How to make AI perfectible, then?
Re: GPT-4
#539Re: GPT-4
#540What I don't understand is how GPT-4 is able to do reasonably well on tests like the AMC12: Many of the AMC12 questions require a number of logical/deductive steps. If GPT-4 is simply trained on a large corpus of text, how is it able to do this? Does this imply that there is some emergent deductive ability that you get simply by learning "language?" Or am I missing something? Obviously, I'm assuming that GPT-4 wasn't…
GPT-4 knew to use linear programming and acknowledged the constraints, even without me formatting the tabular data so the labels were with the values and properly separated! It also ran all of the 2-3 digit integer multiplications/divisions/subtractions/additions correctly. It still failed to "put it all together" in the final step and forgot some constraints. I prompted it "won't I run out of time?" and it acknowledged it then redid it forgetting a different constraint. I wasn't able to get it to come to the right conclusion.
It feels like it has learned a pattern for solving these types of questions but hasn't really gained any actual reasoning about whether it's applying the pattern in a way that makes sense. It confidently announces that it followed all of the constraints when the pattern it chose to follow didn't involve one of the constraints. It then acknowledges it was wrong but doesn't apply reason as much as knows to apply a different pattern that fixes that specific issue.
Another example is I asked it to configure some network interfaces on a Cisco switch in a certain way. I gave it 3 VLANs to configure the interface with knowing 1 was incorrect (in the 5000s, VLANs are only 12 bits long). It created the answer with tagging VLAN 5031. I asked what problems I'd run into running the generated commands and it gave some hypothetical risks, one of which being that VLANs must be in a certain range, but didn't reason that the commands included an invalid VLAN. I told it "isn't VLAN 5031 invalid?" and it apologize and corrected it. I then told it "isn't VLAN 1000 invalid?" and it apologized for it not being a valid VLAN and corrected it all the same even though it was valid.
All that testing the limits said... it may not have emergent deductive ability but I think this learned pattern matching approach based on training situations extends far past where most people would think it would. I think GPT-5 or GPT-6 may well avoid the above problems without necessarily gaining emergent logical reasoning for them as much as just having a larger depth in the patterns.
Large number operations are still interesting though and I'm not sure how they fit in. 646864613385/41348.5 returns "approximately" 15652.172205 which has the right first 3 digits but is off by a factor of 1000 and the rest of the digits are made up. I'm not sure if this is similarly explained by applying a pattern without reasoning about it but it feels like it could be.
All that said I really don't know much about how the system is constructed, I just use it :).