iPad Pro The rise of ChatGPT has made the iPad useless

[iPadified]

But that’s just it.

Sure, the AI was bootstrapped by feeding it data from human pathologists. But by now we not only know that there’s no magic sauce that the humans provide, we know that there is information in the data that the computers can suss out that humans are completely unaware of, almost certainly including features that won’t ever “make sense” to humans.

So what you do is you train the AI not on the information provided by pathologists, but on the actual raw data. You don’t just feed the tagged X-ray images to the AI; you feed it all the medical history that the hospital network has on every patient. In addition to the images, you feed it lab work, you feed it all diagnoses of every condition, you feed it blood pressure, you feed it prescription refill records … you feed it everything.

Of course, the imagery is going to provide most of the information for a diagnosis. But I guarantee you that there’s some cluster of data that’s not on anybody’s radar that also correlates with cancer, and that’s going to let the AI uncover cases that no radiologist would ever notice. And that, in turn, will correlate with some feature on a scan that the radiologist will again be unaware of, so the AI re-reviews all stored images for that feature … which leads to follow-up diagnostics, and so on.

You yourself point to this: “One analysis or opinion is seldom sufficient for large decisions.” And that’s the whole point of “big data”: it automates the process of having teams of experts compare notes.

If you want a simpler case, think of chess. I remember the days when it was taken for granted that no computer would ever beat a grandmaster at chess — and, if it ever happened, it would be a sure sign of human-level intelligence in a computer. We now know otherwise, of course. But today, the thought that you need humans to “validate” a chess computer’s strategy is laughable. The chess computers are having tournaments amongst themselves that no human will ever be able to appreciate.

With ChatGPT passing the bar exam, we’re now at the stage where chess computers were when they started winning individual games in tournaments. They still didn’t have a chance at that time of winning a tournament, but they were running with the pack. Not long after, Deep Blue beat Kasparov. I won’t be surprised if, this time next year — and certainly no more than five years from now — ChatGPT (or some other AI) is consistently cranking out legal opinions that the lawyers who grade the bar exam admit are superior to their own. Will the humans still need to “validate” those machine-generated opinions?

The same will be true of every other academic discipline, sooner rather than later.

So … yeah. Worrying about students having ChatGPT write their essays for them is much ado about nothing. The entire professional class is about to face competition from robots that will make the threat Henry Ford’s assembly line posed to manufacturing jobs seem like a giant nothingburger.

b&

Look up validated and benchmarked against. AI will only be implemented if it is proven superior to humans. We are not there yet in medicine. The difference between chess and medicine is that the board is much larger in the latter so I am less optimistic about an all ruling AI. Furthermore, you need data to feed these systems. Are we gathering the correct data today? Furthermore, my experience from the field is that a bio marker that causes a disease is superior to any combination of proxy biomarker that a AI typically uses. See you own description of noisy data. I discussed the quality of data with a math professor working with AI earlier this week and his conclusion was that the amount of data was less important compared to diagnostic power of the data.

Images, blood values etc are proxy markers, at least for cancer. Cancer is not something strange. There is always a precise molecular mechanism involved and the medicine are developed against these molecular mechanisms.

At some point the AI will run out of compute power. There are 3 Gbases DNA in a human cells, >10 million SNP that modulated cellular processes, about 200 basic cell types (+all the intermediates) where each gene is expressed differently and the level and activity is modulated by SNP that may or may not cross talk. There are >100000 protein expressed at different levels. Furthermore, there are morphological differencies between humans and I am not talking about the face or body mass that influences clinical outcomes. These markers activity are not independent form each other a may cross react and we know not how yet. Furthermore, the immune activity differs between people so it is not an easy equation to solve. Currently cancer diagnostics, we are not measuring all of these markers and we certainly do not understand what each of them are doing. So you see, the board is far larger than chess or the law.

What scares me is the student blindly trust what comes out of a computer. A calculation maybe 6 order of magnitudes off and they still insist the computer is correct because it is a computer. If they do the same with AI we are far from a multifaceted vision needed to push society forward.

 

[dogstar]

Just today I received new instructions from Uni for exams to mitigate ChatGPT: oral exam instead of written exam. Written exam without internet and only physical textbooks as aid (optional). We are even allowed to change exams style in mid term which is unheard of.

ChatGPT just bombed back exams methodology at least 30 years as research based exams are gone for now.

Exams are however easy to solve. Exams will be more expensive but we pass the bill on to the society or the students.

Learning, which is much more important, is a much worse problem because we have no way to track back where ChatGPT got the information from.

Who is happy with this progression? Students?

I mean, ultimately I think tools like ChatGPT can be useful for research, just like textbooks, google search, etc, but man it is critical students learn how to critically think for themselves or I think we are eventually done for as a society. I almost think the best ChatGPT type interface will incorporate search results with its text so users still realize ChatGPT is just a sort of summary and often is incorrect to a degree and it's always better ultimately to go to the source for the information you need to verify it if it's important.

 

[LionTeeth]

I don’t know what “ChatGPT” is, and my iPad works perfect. Same as it ever has. I don’t understand how an “if/then” piece of software can replace a hardware unit like the iPad.

 

[adib]

I've never heard of anyone describe it as "hallucinating" but that's really hitting the nail on the head. I've been impressed with when ChatGPT is helpful but it has completely made up answers and given me sample code that would not compile several times.

FWIW, Google is not a source of "factual" information. Neither is Bing.

 

[Solomani]

Furthermore, it does not seem to say from where it takes the data/knowledge so it is in that sense quite useless without significant literature study where iPad is outstanding for reading books and scientific articles.

University Dean: "Until further notice, you are on academic suspension for plagiarism."


Student: "But but but Sir! I did not plagiarize. It was the ChatGPT bot that forgot to cite the sources for my research paper!"


University Dean: "Just get out of my office... NOW!"

 

[MisterSavage]

FWIW, Google is not a source of "factual" information. Neither is Bing.

Well I was referring to coding questions and I use Google constantly for that.

 

[TheRealAlex]

I have witnessed that people in my university, and a lot of people on social media, have adopted ChatGPT into their workflow. These "co-pilots" are a crucial part of working and they live permanently on the side. You are constantly sending it ideas and editing the draft. For that to happen, you need a keyboard which makes the iPad without a keyboard useless for people working in 2023. Of course, people say you can use the iPad with a keyboard but at that point, you might as well use the MacBook instead.

What are your thoughts on iPad's place in a world of AI butlers, like ChatGPT?

ChatGPT has revolutionized my projects. Just in extra income because I have a co-pilot now. It’s an extra $3,200 dollars a week. And I work less.

 

[Rafterman]

ChatGPT has revolutionized my projects. Just in extra income because I have a co-pilot now. It’s an extra $3,200 dollars a week. And I work less.

Be careful. ChatGPT may someday replace you entirely.

 

[iPadified]

ChatGPT has revolutionized my projects. Just in extra income because I have a co-pilot now. It’s an extra $3,200 dollars a week. And I work less.

What do you do for work?

 

[SoldOnApple]

I use iPad as a drawing tablet. Though now long until OpenAI comes up with something that ruins the career of aspiring illustrators...

 

[TheRealAlex]

What do you do for work?

Myself. And I’d rather keep my projects private so less people tap in to the revenue stream.

 

[floral]

Myself. And I’d rather keep my projects private so less people tap in to the revenue stream.

Tap in? You're making this sound less like "I'm using a publicly available and extremely popular AI service to deceive people into thinking that my work is human" and more like something you'd find on the dark web.

 

[iPadified]

Myself. And I’d rather keep my projects private so less people tap in to the revenue stream.

Without substantiating your claim and verifying it, your statement is difficult to believe in 100%. You might be right or you might not be, how could I tell? Just like some ChatGPT "claims" or any internet source.

 

[TheRealAlex]

Without substantiating your claim and verifying it, your statement is difficult to believe in 100%. You might be right or you might not be, how could I tell? Just like some ChatGPT "claims" or any internet source.

For all you know my responses could be ChatGPT, and not from a real person.

 

[TheRealAlex]

Without substantiating your claim and verifying it, your statement is difficult to believe in 100%. You might be right or you might not be, how could I tell? Just like some ChatGPT "claims" or any internet source.

yeah I’m not those get rich quick schemes telling people what to do. I’d rather NOT tell and keep it all to myself.

 

[Rafterman]

South Park's last episodes' credits claimed it was co-written with ChatGPT.

 

[Solomani]

South Park's last episodes' credits claimed it was co-written with ChatGPT.

Good. They can fire all their writers and producers now.

Give the job to the bots.

 

[xaqt93]

I have witnessed that people in my university, and a lot of people on social media, have adopted ChatGPT into their workflow. These "co-pilots" are a crucial part of working and they live permanently on the side. You are constantly sending it ideas and editing the draft. For that to happen, you need a keyboard which makes the iPad without a keyboard useless for people working in 2023. Of course, people say you can use the iPad with a keyboard but at that point, you might as well use the MacBook instead.

What are your thoughts on iPad's place in a world of AI butlers, like ChatGPT?

I adisagee with your statement. I don’t own a MacBook. I use my iPad Pro with Stage manager on an external display. I have a keyboard and mouse. When I am on the go, I use the iPad like a tablet. I also use it for flying as I am a pilot and find it very useful in lots of instigation with my job from flight planning (with or without a keyboard) and my day to day. I even use Photoshop on it. So far, I have not found a single thing in my day to day life that I was like, “I need a Mac for this” or thinking, “if only I had a MacBook.” My iPad can do everything I need it to do. For most people, an iPad is going to be great.

The only people complaint about the iPad not doing enough is the people that do really intense work like developers and film makers, even some YouTubers get away only using the iPad with Luma Fusion and other video editing options that are out in the world. When it comes to the basic tasks of a computer that an average consumer would need, iPad can do it. Just my opinion

 

[Retskrad]

Maybe I was expecting too much from the iPad Pro. When I use my M2 iPad Pro with the Magic Keyboard, I feel very inefficient and everything takes longer than it should. It’s not suitable at all for intense intense tasks, like university studies where you do a lot of research and jumping between different windows while simultaneously going back and forth with a co-pilot like ChatGPT.

The iPad excels with the Pencil, consuming content and other light tasks. Maybe we should think of the iPad as an accessory to the Mac and not a standalone thing just yet. Apple tried to market the Magic Keyboard as something that makes the iPad a laptop replacement but for a $400 keyboard it barely adds anything to the iPad experience. It’s just a keyboard. iPadOS is a tablet first experience and that’s how it should be used IMO.