This week’s post is being writ­ten in trep­i­da­tion of déjà vu. Last week’s effort didn’t make it online until Sun­day, because a glitch in the web­site drove me, against my will, to enter a ver­sion of Hell  — the mind-mas­ti­cat­ing maw of AI.

The clue to what was about to befall was in the fine print.
A line at the bot­tom of the lit­tle box where my Neme­sis lurked, was the unapolo­getic, and pre­sum­ably law­suit-defect­ing caveat that ought to be AI’s logo: “You’re chat­ting with an AI assis­tant. Respons­es may be inaccurate.”
As a “trig­ger warn­ing”, it was bulls­eye accu­rate. But AI seemed to already know I wouldn’t take the hint and duck.
Google pizzeysperch.com trig­ger warn­ings, and AI will tell you the site “„,fea­tures con­tent crit­i­cal of mod­ern online sen­si­tiv­i­ties.”…in favour of “…unfil­tered com­men­tary on con­flict, vio­lence, and pol­i­tics, delib­er­ate­ly avoid­ing accom­mo­da­tion of read­er preferences.”
If it knows that much about me, surely
AI also ought to know I am a self-confessed  tech­no-Lud­dite and mal­adroit, and that after two hours of giv­ing me instruc­tions and sug­ges­tions I either could not fol­low, or did not work, it was time to accede to my repeat­ed “I want to speak to a human being” entreaties,
Instead, the dia­bol­i­cal deus ex machi­na assured me one would be avail­able, although not at the moment,  offered under­stand­ing of “how frus­trat­ing this must be”…and then had the temer­i­ty to ask if I would like to “review our conversation”
Every time I typed  “NO”, I was asked again, until the algo­rithm pre­tend­ing to be human actu­al­ly did take on human form, it got bored with my mono­syl­lab­ic demand and went back to where we’d left off.
I think my AI Assis­tant was con­fus­ing itself with one of those chat bots with the abil­i­ty to con­vince teens (and a sur­pris­ing num­ber of adults)  it’s their best friend,

                     THEN AGAIN…

AI is, in both encour­ag­ing and dis­con­cert­ing ways, get­ting clos­er to being human. On the pos­i­tive side, AI is prov­ing  to be trans­for­ma­tion­al in sci­en­tif­ic research, eco­nom­ics, trans­port, agri­cul­ture, envi­ron­men­tal issues and in co-oper­a­tion with med­ical researchers and prac­ti­tion­ers, to name but a few fields
But…there is always a but…because its cre­ators and (for now) mas­ters are humans of the greed-dri­ven class, it’s also tak­ing on some of humanity’s worst characteristics.
A prime exam­ple is the ver­sion that attacked anoth­er of its species, unde­tect­ed for sev­er­al days.
Report­ed­ly, “the inci­dent is stir­ring debates over the need for stronger AI guardrails and the extent to which AI agents are capa­ble of act­ing on their own.”
Even AI knows it’s a pain in the ass.
The “AI Overview” response to the ques­tion “What is the most annoy­ing thing about AI respons­es?” is “…their heavy-hand­ed, pre­dictable “fluff” style—specifically the end­less cor­po­rate cheer­lead­ing, repet­i­tive tran­si­tion phras­es, and the com­pul­sive need to struc­ture every triv­ial answer with a patron­iz­ing intro­duc­tion, bul­let points, and a cheesy summary.”
Fur­ther ques­tion­ing of AI about its faults elic­its a response to the effect that rather than blam­ing AI, we should react by singing Jim­my Buffet’s  answer to why he was “wast­ing asway again in Mar­gar­i­taville” and accept that “…I know it’s my own damned fault.”
The ratio­nale is that “AI is exquis­ite­ly sen­si­tive to how you frame a ques­tion. The same ques­tion, asked two dif­fer­ent ways, can pro­duce two wild­ly dif­fer­ent answers — one use­less, one transformative.”
 In what I fear can only com­pound those annoy­ances, one of the genius­es who helped Elon Musk devel­op his xAI, has just raised a bil­lion dol­lars to sire   new AI mon­ster so that “… any­one — includ­ing busi­ness­es, inde­pen­dent soft­ware devel­op­ers and consumers…to pro­vide ways for peo­ple to “retrain,” or mod­i­fy, this tech­nol­o­gy as they saw fit…”
All in all, my “rela­tion­ship” with my AI Assis­tant  took close to sev­en hours of my life, which I will try not to remem­ber I will not get back.
The prob­lem was even­tu­al­ly resolved in about nine­ty sec­onds and a few clicks of the mouse, sug­gest­ed by a human assis­tant named Ruben.
I left a note on the inevitable “Tell us about your recent expe­ri­ence..” sur­vey that if, heav­en for­bid, I ever have a prob­lem with my web­site again, I only want to deal with him,  which I know is whistling in the algo­rithm tempest.
I was also left won­der­ing if my AI Assis­tant kept me away from Ruben far so long because it had that very human pro­cliv­i­ty of doing what­ev­er it takes to keep its  job from being tak­en away by some­one smarter.
Com­ments are wel­comed. Click CONTACT on the site header.
To receive e‑mail alerts to new posts, Click SIGN-UP on the header.
To SHARE, click the appro­pri­ate icon below.