Posts by Reiki Shangle

...and an inquiry too?

What's going on with Sam Altman's voice?

Has he been to an elocution coach to help him perfect a slow, low-pitch growling drawl?

Has GPTx advised on how to best inflect and converse?

We should be told.

Hallucination aka Wrong

I prefer not to anthropomorphise and call this problem Hallucination. Instead I keep things simple and call the results “wrong”.

What a mess…

The first rule of AI is…

…don’t drink the Kool-Aid.

Any technology which is based on training from human-sourced data, which requires so much supervision or checking due to human-like mistakes, and where in a recent Gemini chat exclaimed that it couldn’t check the spelling of words to ensure that they ended in ‘ity’, is perhaps better known as a Pig in a Poke.

Clearly. gullible doesn’t exist in the AI dictionary.

1bn?

1bn or more might get a positive response, and still provide a lucrative investment for both the country and the company, whilst keeping competition on their toes.

Preemptive strikes with Liberationware (TM)

1. Find points of entry. Zero-day, social, technical, insider, etc.

2. Install firewalls, close ports, update passwords after installing a password manager and communicate access to known trusted parties.

3. Inconvenient? Perhaps, it’s a pain to have to use a password manager. But is it? I’ve used 1Password for more than a decade and it’s essential, simple, and it even makes Passkeys usable.

Ransomware problems stem from organisation and training failings. There’s no excuse for treating confidential customer and patient data like this.

And don’t start me on the scam that’s called The Insurance Industry.

Simon says…

…make them bigger, so they make them bigger.

Simon says, pause and think. So they just ignore Simon and plough ahead, armed with bigger, and riskier.

Simon says, weep and despair. But sees we beat him to it.

Simon says… ¡jesu cristo y la virgen, que cojones hacéis! as a way to bypass English profanity filters, whilst concisely expressing the disdain and despair.

Organised, moi?

People don’t like to organise anything. Their data, their cupboards, their meetings, their life. Nothing.

Anecdotal, perhaps, but I must be special to have been selected to meet the majority of them…

Gearing up for…

This sounds like the service providers are gearing up for the largest game of wordwang*, because they forgot to take their large dose of reality.

Have all other problems now been subsumed and assumed fixed by Alt-Man et al? I hope not. The lecture notes are woeful and I don’t see any useful, credible examinations, as you would expect of any expert.

Of course, Alt-Man ads might just find a use for this dubious quality data. At the end of the day, whose wallets are going to bulge? And what else can these GPUs be used for?

I guess with the declining relevant working populations, there is now a rush to get some infrastructure in place before there’s no-one trained locally to do basic tasks.

I’ll let myself out…

* Mitchell, Webb et al.

Déjà vu…

Intel, the British Leyland of the chip industry…

Mixed and contradictory messaging, is that a new thing?

Corporate messaging and legals that need to be “confirmed“ or “clarified” through a global fact-free chit-chat network (Xitter) are a sure sign that chaos and the well-worn try-it-and-see-what-happens management stance is determining direction and public relations strategies. I wouldn’t expect such unprofessional work from an employee, and if my new AI substitute employee exhibited such behaviour, I would want to take advantage of the return policy, or I’d call my insurance company to manage the mess, whilst reviewing my newly minted AI job-replacement program. Repeated ill-considered behaviour doesn’t indicate that things will improve.

What a mess.

William Hill Consulting?

Did they really retain McKinsey & Company, or did they actually bring William Hill on-board?

Personally I don’t rate the odds on the gambles being taken here and elsewhere. Risky strategies based on unverifiable incantations.

Whatever next? The Nostradamus LLM?

Bush prophesy

Experiment IV edges closer…

Free as in, of no value…

At least Broadcom are making their intentions and actions clear. In a crowded virtual and container market, I doubt I’ll use Fusion again. They’re clear about the value of the product to them; zero. The VMware business strategy was never clear, never ambitious nor extensive, sometimes weird (I’m talking about the RabbitMQ acquisition, which is a great product, but surely a sore thumb), and so much potential was squandered.

At least they’re not Google.

Google missteps

Oh jeezuz.

We have Bernie-Lee to thank for these glorified, excessive infostructure systems.

Now, can AI help me fill this form for my Rail Pass renewal whilst the Javascript assumes I want to move automatically to the next field of my Passport numbers? Can the AI take over the UI testing and forward the expletive-filled results to the corresponding AI dev?

Thanks. Thanks a lot.

Barn door policies

I guess I sit in the company of only a few, to harbour the view that there must be a hell of a lot of horse roaming outside those damn barns, and it must be a profitable business for someone or some group solely for the lack of concern or understanding of risk, and budgets that are rarely concerned with risk because it often evades having a specific value.

But what do I know. (Opens browser to check if insurance company will offer discounts based on my security systems and processes, backups, and staff training.)