Somehow, Microsoft and OpenAI have succeeded in dominating the internet from news to blogs in a way that would have seemed unimaginable had we not seen it. Bear in mind that the adversary is Google, the king of the content castle. Tis is like selling hamburgers to McDonalds staff inside their stores and yet we can’t login without another gusher about how amazing chatGPT was and now GPT4.
- What we are seeing is smoke and mirrors, you now have a better likelihood than ever before of a convincing looking response to a question, were it allowed to do so. That’s because of ML and NLP and years of munging big training datasets. The computer is no smarter today than it was in 1975. To be blunt about it, a computer in 22023 can discern a 0 form a 1 or rather on form off. Just like it could then. Another day, Ill enter a debate about whether humans are any smarter.
- This is not from a blog about an article about an essay about a paper about AI.
This is easy to follow and not complicated.
chatGPT, the first semi-intelligent bot that despite its ‘genius’, couldn’t answer a question like; “Why can’t I log into my Onedrive account for the past two days?” was moved instead to a side show where the medicine mans freckled daughter and her ugly uncle keep asking the bot apparently smart questions to make it look good, you know, the sort you asked it yourself because you’d seen the medicine mans ugly uncle asking it and getting an apparently great answer.
I’ll point out to you with confidence that you didn’t check that answer properly and you didn’t give it a fair test of how likely you are to have it write your essay and submit it automatically on your behalf. - Here’s what happened to me, stay safe please.
For two days I have been unable to continue a writing project because my content is in onedrive and both my laptop and desktop stopped communicating with it. As you’ll be aware I no longer have much say about using onedrive, it’s that are go to Ubuntu, (another post coming soon)
As best I can gather, it doesn’t recognise most of the time that an account exists for my Microsoft email address and refuses to accept any password
Occasionally it lets me move on to sign in (presumably recognising that an account exists) but then refuses my password.
Once in desperation, after many attempts it sent me a code and I was able to reset my password. I thought yippee. No
it asked me to log in and then turned down my new password.
On the next attempt it said the account didn’t exist.
There’s an article somewhere online mentioning a confusion about old outlook accounts and live accounts throwing the whole login system at Microsoft in a tizzy and some very longwinded instructions whereby they reckon a mere mortal might eventually get past the monster and get their document. I don’t know where to find it again and I’ve decided to give up.
A. You cant call their phone number any more
B. The one to one help by chat no longer exists.
C There isn’t even a dimwit bot making a reasonable attempt to ask you what’s wrong.
d. There’s just a small number of options, the ones MS are prepared to try and fix and if it’s not those, well tough luck. We’re a monopoly after all.
Too bad if you’re going out of business, or you can’t submit your paper, or your article is about to miss deadline, or the court will lock you up for not responding by two hours ago, or all the other real-life problems that won’t wait for Microsoft intelligence.
4. Why it is such extraordinary hypocrisy.
Imagine, just imagine if Microsoft had access to an intelligent agent that could use a bit of logic and help stranded, hard-working, honest customers who’ve paid their subscription and just want a forking life!
5. What GPT4 actually amounts to.
a. Better plagiarism.
Twenty years ago, I wrote NLP to help academics along with lots of other bright programmers. We parsed documents, pulled out the key concepts and then tagged them, building an Ontology as we went. With tiny budget it went very well and very far. One of the tricks I achieved was to tell it to create a website about “topic” and it created perfect IA and a great resource, just for you. That didn’t take billions. What they’re doing isn’t incredible and if you ask the embarrassing question about what it cost? Who cares when the military are dumping money into it. Google translate was the first time we say a system look for a concept and recognise it in a different language, not just different style and choice of words, but anther language. It’s a similar approach and once you recognise that, understanding how chat GPT can paraphrase my article to sound like I like Microsoft isn’t really that hard. That’s all you are seeing. Mass processing, caron emissions on a huge scale and nothing but horse manure to show for it.
b. Smarter google without a revenue model.
There is most likely an obvious reason why Google wont apply a similar approach to search, despite the fact that tit would work very much better and in reality a smart search engine is precisely what it is.
“Google sells the most profitable answer to a profitable question” Ed Taaffe ( I like Google too)
If there’s no money in describing the mating rituals of the bumble bee, it won’t make the top of the SERP (results page) for your search. You might see expensive honey or busy Bees recruitment agency or whatever, but no bumble bee sex for you.
Why? Well there’s no money in it, now is there?
Ask yourself this and if you have an inspiration, please contribute the discussion because I’m lost on this one. Why would these companies take such a risky approach to churning up interest for an undeveloped product that they cant find a revenue model for?
Apart form spreading disinformation for the people who financed them, what else are they good for for the next few years?