LegalOpsHezzle @LegalOpsHezzle

Legal Ops, Tech & Strategy In-House @ adidas. Commercial Lawyer. Views my own and RT ≠ endorsements. #LegalTech #LegalDesign #LegalInnovation Manchester, England Joined January 2021

Tweets

368
Followers

940
Following

2K
Likes

1K

LegalOpsHezzle @LegalOpsHezzle

a year ago

@scottastevenson I explained vibe coding to some lawyers from one of my teams in the pub yesterday. I don’t think they’ll invite me next week.

2 0 5 119 0

View Details

LegalOpsHezzle @LegalOpsHezzle

2 years ago

@scottastevenson I like the idea of comparing to athletic fundamentals that are essential standards and abilities needed to be in the game at the requisite level you want to compete at, and then rest is your special skills and abilities that allow you to play a certain way.

0 0 1 33 0

View Details

LegalOpsHezzle @LegalOpsHezzle

2 years ago

@ItsMattsLaw

0 0 1 83 0

View Details

LegalOpsHezzle @LegalOpsHezzle

2 years ago

1 1 8 296 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@_M155Y_ @BBCWatchdog @British_Airways Dreadful - i've been waiting since September, but when I check Twitter periodically I can see people chasing up their claims from 6+ months ago. Same old replies. They're prime for an investigation by BBC and by the regulator. It's shocking service. I think I'll just sue in 2024.

0 0 0 54 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@ZachAbramowitz GenAI tools are in my daily arsenal of tools and save me lots of time on certain tasks. I was just interested if you had some specific examples of what you thought they were great at? (I hate twitter for always reading cold and sarcastic - I’m genuinely interested)

2 0 1 57 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@jackwshepherd Enjoyed this - really nicely laid out. The 7 capabilities scribbled down for reference.

0 0 0 63 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@sobedominik Whoa.

0 0 0 112 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@SkyNews Typical. You pray for one religion and three turn up at once.

0 0 0 181 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@Nicola_Shaver Oh dear my ChatGPT effort kicked this out…maybe it was my prompt “Draw me a picture of a thanksgiving turkey being stuffed with legal tech products / apps by a maniacal woman obsessed with both legal tech and turkeys”

1 1 5 208 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

At the risk sharing a GenAI first world problem, but god I'm sick to death of waiting for the response to be typed out.

0 0 0 149 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

Is anyone here using Co-Pilot in the legal space yet - do you have a standout use case that you've adopted out of the box? Got my licence today.

2 0 0 250 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@HBurness @BamLegal @alexgsmith @Nicola_Shaver @nickyleijtens Logo on a power bank. Done!

0 0 0 105 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@damienriehl @ReenaSenGupta Interesting. Damien / Reena any data for the corp law depts lagging on GenAI behind law firms comment or more anecdotal?

1 0 0 34 0

View Details

Ara Ghougassian @araghougassian

3 years ago

your kidney can go for like $300k and you only need one to survive god gave us all startup capital. you just have to want it bad enough

248 1K 8K 1.1M 418

View Details

The_AI_Skeptic @The_AI_Skeptic

3 years ago

Long Post: Why Generative AI is Currently Doomed We're so used to technology getting better. Every year there's a new iPhone with a faster processor. It's the way of the world... or so it seems. Sometimes, bigger doesn't mean better. Take, for example, LLMs like ChatGPT. If you keep scaling them up, they eventually become worse. This inverse scaling leads to them becoming actively bad: (youtube.com/watch?v=viJt_D…) (When @OpenAI were developing the highly anticipated ChatGPT-4, it seems they may have hit this problem already. Because instead of scaling up their training sets, like previous iterations of their model, leaks indicate that ChatGPT-4 may actually be 8 x ChatGPT3 models tethered together (thealgorithmicbridge.substack.com/p/gpt-4s-secre…), explaining why it was delayed and why the dataset size was not revealed.) But even if you believe they'll find a way around that problem, there's still plenty of others waiting for us. What if I told you that language model technology isn't actually new? What if I told you it's largely the same as it was in the 1980s, but the only thing that's changed is the transformer technology, allowing for more efficient training, and the sheer size of the training data: The public internet. Yes, the thing that gives ChatGPT (and other LLMs) their "magic" is the fact that the internet exists now and can be scraped. (After it's being manually catalogued by hundreds of thousands of foreign workers, of course (theglobeandmail.com/business/artic…).) 300 billions words from the internet were used to train ChatGPT-4. It's the scale of this training dataset that allows it to sound so human and knowledgable. There's nothing else like it. Not only are major companies preventing their content from being used in future AI training datasets (deadline.com/2023/10/bbc-wi…) but there's a lingering question on whether or not it was even legal for them to use their original datasets in the first place (theverge.com/2023/9/11/2386…). But worse than that, the internet is increasingly being polluted with error-ridden AI generated content. So much so that it's infecting search results (wired.com/story/fast-for…) And in models that talk to the internet, these mistakes are now being fed back to us through AIs (futurism.com/google-search-…). (m.facebook.com/story.php?stor…). Sometimes even sharing conspiracy theories as fact, oops (washingtonpost.com/technology/202…). And now that AI generated content is everywhere, that means it's in the next training dataset. Except you can't train AI on AI generated content (arxiv.org/pdf/2307.01850…). And no, you can't reliably detect AI generated content either (x.com/atoosachegini/…). So, even if bigger meant better, and there wasn't an inverse scaling issue, where is the next dataset going to come from? How can we keep AIs up to date if they are polluting their own learning pool? There is no other conclusion than the future looks bleak for generative AI...

Atoosa Chegini @AtoosaChegini

3 years ago

Check our paper! Our study reveals the unreliability of current image watermarking methods in AI-generated image detection. We've also shown the theoretical impossibility of achieving both reliability and imperceptibility in watermarking methods. Link: huggingface.co/papers/2310.00…

2 7 55 29K 19

0 12 46 9K 33

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@SavageFridays Cheers, Jenn - enjoyed the live CLOC Talk at the end!

0 0 0 16 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

Here are my paraphrased notes from Jim Wagner's comments in the "Generative AI: How to Find the Perfect Fit" session. It was an interesting session with some other speakers (hence notes jumping around topics):

LegalOpsHezzle @LegalOpsHezzle

3 years ago

At #CLOC London today. Very heavy GenAI speaker list…interested to see where the circuit is at with it all.

2 0 5 2K 1

2 2 4 1K 3

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@nwaisb Maybe there might be more success for vendors in the second wave, where other inspired orgs / competitors are looking for something similar but can sell it a bit more out of the box in a more scalable and sustainable way.

1 0 1 29 0

View Details

LegalOpsHezzle @LegalOpsHezzle

3 years ago

@nwaisb These phases seemingly being done with a sandbox GenAI tool, consultants and SMEs building and testing on the fly. No real space for tool vendors unless they've already hit on and have successfully marketed their solution in that very specific space.