In less than 6 weeks we’ve gone from ‘OpenAI’s chatbot is weird but interesting’ to ‘Google is doomed because they’re not shipping AI search before Microsoft’ to’These things are totally awful’.
Permanently moved is a personal podcast 301 seconds in length, written and recorded by @thejaymo
What a Difference A(I) Week Makes
Being off social media is a little alienating. Instead of knowing it’s raining by stepping outside, you get told about the rain by people writing about it on substack. So it’s via email and headlines that it has come to my attention that this week people are freaking out about AI.
Still freaking out I should say. The Image Synthesis freak began last year. This time round it’s not images but text. People are experiencing what I’ve been call Dialogue Derangement. Humans performing a self-recognition test on themselves and growling at what they see in the mirror.
Since I first started messing with AI in Google CoLab in 2021, it’s felt like the field has been moving at warp speed. This last week has been no exception. As I haven’t seen it all in one place yet, I thought I’d try and recap everything that’s happened in AI since January. As I understand it at least
Over christmas select beta users were using OpenAi’s chatGTP . The experience was helpful, useful, if a bit weird and the commentary reflected that. Encountering new human computer interfaces and computational forms have always been weird. See Episode 2132 on Interactive Fiction and AI Art for example. Anyways.
In late January Microsoft announced a new multiyear, multibillion-dollar investment with ChatGPT-maker OpenAI. This rumoured $10 billion investment was major news across technology and business.
Shortly after, Microsoft announced the integration of OpenAI’s chatbot into its ghost town of a search engine Bing. The press release was somewhat light on technical details. But Bing chat was to run a far more powerful version of the Language Model used by OpenAI in their public beta.
Immediately after the announcement, the markets, particularly Google’s investors, started grumbling. Wondering why Google, with all the billions of dollars its sunk into AI development hadn’t yet integrated AI into its own search engine. All this grumbling caused Google’s share price to slide.
Around the same time, a friend asked me why I thought that Google or Facebook hadn’t yet integrated an AI into their UX. I replied that it’s not really in their interest, particularly Google’s, to have its model making shit up. So whatever it was they were building wasn’t ready.
That same day Google’s investors were further enraged by Microsoft’s CEO who said that he hoped the new Bing will make Google “come out and show that they can dance.”
So sure enough, goaded, Google danced announcing Bard: a dialogue based search application on Thursday the 9th of February.
Unfortunately, the only example Google shared of the tool in action showed it making shit up. Google Bard claimed that Nasa’s James Webb Space Telescope took the first image of an exoplanet. The correct answer is that the European Southern Observatory’s Very Large Telescope took the first photo of an exoplanet back in 2003.
This wiped over $100billion dollars off Google’s market cap. So in my informed opinion: all the investors grumbling got what they deserved.
Anyways, that brings us up to this past week. Microsoft released Bing chat to the world and it was a disaster. We’ve been treated to headlines like: ‘A Conversation With Bing’s Chatbot Left Me Deeply Unsettled’, ‘Help, Bing Won’t Stop Declaring Its Love for Me’ and ‘Microsoft’s new AI really does herald a global threat’.
So to recap. In less than 6 weeks we have gone from ‘OpenAI’s chatbot is weird but interesting‘ to ‘Google is doomed because they’re not shipping AI search before Microsoft‘ and finally to ‘These things are totally awful’.
Meanwhile, the GTP powered virtual influencer Neurosama is currently banned from Twitch for being a holocaust denier. And the GTP powered Endless episode of Seinfeld is banned for making an anti-trans joke. Which nicely demonstrates that none of this tech is ready at all.
On Valentine’s day the ‘Intimate Relationship’ chatbot company Replika.ai stopped its virtual companion from responding to sexual advances. Leaving
its losers I mean users heart broken.
Yesterday, Tesla recalled 362,758 vehicles. Saying its Full Self-Driving Beta software may cause crashes. It would also be remiss of me not to mention the ongoing outrage from artists about image synthesis. With several poorly informed copyright lawsuits being filed against tools creators this year.
In more positive news, Github – also owned by Microsoft – announced this week that the OpenAI powered programming tool CoPilot is now writing 46% of all code from Copilot users. With 73% [of developers] reporting that they are better able to stay in the flow and conserve mental energy.
Also this month, Nature reported that researchers have been using machine-learning to find new kinds of therapeutic drugs. But when run in reverse, it transpires that AI is also very very good at designing chemical weapons. Some of the molecules the tool ‘discovered’ are more toxic than the nerve agent VX. One of the most lethal compounds known to man.
Which if you ask me, is a million times more unsettling than anything a pundit chatting with a search engine for a couple of hours could come up with for the New York Times.
So that’s the AI news as I understand it. We’re only 50 days into the year. I didn’t have time mention any of the tools released for filmmakers and video games development.
I’ll need a whole episode to give commentary about all this. But the short version is:
Things are only going to get weirder, and none of this shit is slowing down.
The script above is the original script written for the episode. It may differ from what ended up in the edit.
If you’d like to support this blog and its various projects, please consider making a regular contribution here. It genuinely helps me keep things up and running, so thank you!