The Present is GPT‑3: The Future?

“I had not real­ized … that extreme­ly short expo­sures to a rel­a­tive­ly sim­ple com­put­er pro­gram could induce pow­er­ful delu­sion­al think­ing in quite nor­mal peo­ple”. — Weizen­baum, 1976.

As anoth­er impor­tant step in the devel­op­ment of arti­fi­cial intel­li­gence, peo­ple are now talk­ing about Gen­er­a­tive Pre-trained Trans­former 3 (GPT‑3) because it is way bet­ter than any lan­guage pro­gram in exis­tence. It can pro­duce text that any human can read. And this break­through can be crit­i­cal for com­pa­nies that wish to auto­mate most tasks.

GPT‑3 can con­tex­tu­al­ly respond to text inputs. For exam­ple, com­pa­nies can use it to enhance con­sumer ser­vice with­out giv­ing cus­tomers the feel­ing that they are chat­ting to a machine.

Sev­er­al down­sides come along with huge poten­tial. GPT‑3 is a deep neur­al net­work that is still large­ly unknown to humans. Its algo­rithms can­not be exam­ined or stud­ied to under­stand how they work.

Some claim that while the text that GPT‑3 writes first seems great, the sys­tem’s lan­guage los­es coher­ence and becomes illog­i­cal when work­ing on longer works.

Many peo­ple are also con­cerned that GPT-3’s fail­ure to dis­tin­guish between fact and fic­tion can be used to pro­mote social bias­es like sex­ism and racism. For instance, GPT‑2 was not avail­able to the gen­er­al pub­lic since it could be exploit­ed for spam­ming and the spread of mis­in­for­ma­tion.

Can GPT‑3, which is more advanced than GPT‑2, have a big­ger poten­tial for abuse and mis­use?  Pro­po­nents of the algo­rithm are required to answer such questions.

GPT‑3 vs Human Intelligence

In a recent­ly con­duct­ed exper­i­ment, GPT‑3 AI Suc­cess­ful­ly Imper­son­at­ed Famous Philoso­pher Daniel Den­nett. Philoso­phers Eric Schwitzgebel, Anna Strass­er, and Matthew Crosby’s exper­i­ment asked peo­ple to choose from 5 answers.

Each ques­tion had five options to choose from; one was by Daniel and the remain­ing 4 were gen­er­at­ed by GPT‑3.

It was per­fect­ly indis­tin­guish­able for the gen­er­al pub­lic, and con­fus­ing for the experts and expe­ri­enced blog readers.

The exper­i­ment involved 98 online par­tic­i­pants from Pro­lif­ic, 302 peo­ple who clicked on the quiz on Schwitzge­bel’s blog, and 25 indi­vid­u­als with in-depth famil­iar­i­ty with Den­net­t’s work who were con­tact­ed by Den­nett or Strass­er.

Accord­ing to Schwitzgebel, they expect­ed the Den­nett experts to cor­rect­ly answer at least 80% of the ques­tions on aver­age, but they only obtained a score of 5.1 out of 10. No one cor­rect­ly answered all ten ques­tions, and only one per­son got nine. The aver­age accu­ra­cy rate among blog read­ers was 4.8 out of 10. The top­ic of whether humans might con­struct a robot with beliefs and desires bewil­dered the spe­cial­ists the most.

Since its intro­duc­tion, GPT‑3 has spawned dozens of addi­tion­al exper­i­ments that lead to sim­i­lar eye­brow-rais­ing reac­tions. With very lit­tle prompt­ing, it can gen­er­ate tweets, write poet­ry, sum­ma­rize emails, answer to triv­ia ques­tions, trans­late lan­guages, and even writes its own com­put­er pro­grams.

It seems that GPT‑3 has not only made a strong­hold at present in arti­fi­cial intel­li­gence (AI) tech­nol­o­gy, it also indi­cates a bet­ter future ver­sion of the same. Whether the next gen­er­a­tion of GPT will be able to beat human intel­li­gence or not is still unclear. How­ev­er, it’s already mak­ing con­sid­er­able head­way and is one of the most fas­ci­nat­ing steps in the his­to­ry of AI.

GPT‑3 representing “the present”

As a con­ven­tion­al log­ic, we can say that GPT‑3 is a huge break­through in the realm of AI. The sin­gu­lar­i­ty is almost here; yes. dig­i­tal sin­gu­lar­i­ty. AI will be able to do many things that humans can do and maybe even some things that we can­not…imag­ine!

Aside from the fact, it’s so smart that there is not much room for a human to under­stand its decisions.

As long as algo­rithms are trapped inside the machine, there is no rea­son to wor­ry. As for the GPT, it won’t devel­op its own agen­da. But that doesn’t mean that we should not wor­ry to pro­tect our free­dom and integrity.

With deep learn­ing, we cre­at­ed ‘black box­es’, just like GPT‑3. So with that said, we must decide how to use them. GPT‑3 is a great tool for us to know what is not good with AI and maybe how we can improve our­selves if we can’t be self-aware with­out it.

We know that no tech­nol­o­gy can remain as it is for­ev­er. GPT‑3 will also go through trans­for­ma­tion, ren­o­va­tion and fur­ther devel­op­ment along with time. When its next ver­sion is intro­duced, there won’t be any sur­prise if it could move for­ward and pre­fer to use advanced meth­ods instead of man­u­al data entry.

GPT‑3 is not the future — it’s the present. We as well as oth­er advanced algo­rithms, includ­ing GPT‑4, are the future! As GPT‑3 is just the always chang­ing present, we need to be ready for that change in every­thing where we can.

Read:Our Future with Virtual Artificial Intelligence(VAI)

Here, being ready for the change means to pre­pare our­selves to deal with the poten­tial impacts of change. We can’t judge GPT‑3’s abil­i­ties as neg­a­tive or pos­i­tive, but we need to be aware whether it and its oth­er ver­sions are doing good things or bad in the field of technology.

We should also see how GPT‑3 itself is evolv­ing with or with­out human inter­ven­tion. It means that we should be able to see and pre­dict its evo­lu­tion with or with­out our con­trol in the near future.

Significance of GPT‑3 at Present

Humans are good at under­stand­ing what they see, and we are par­tic­u­lar­ly adept at under­stand­ing sen­tences. The rules are fair­ly straight for­ward: you look at a sequence of words, you pause to think what they might mean (or look them up in a dic­tio­nary) — then you try to work out how they fit togeth­er and how this impacts your state of knowledge.

When these skills break down, as they can do in some forms of autism or lan­guage dis­or­der, it can be very dif­fi­cult for peo­ple to under­stand what’s being com­mu­ni­cat­ed to them. In such, GPT‑3 is the best tool that can sim­pli­fy our life.

GPT‑3 is a sig­nif­i­cant break­through, and it already has plen­ty of appli­ca­tions. It has been able to gen­er­ate accu­rate text in dozens of lan­guages, com­plete with diverse accents and styles. GPT‑3 per­formed ahead of humans in every aspect of com­pre­hen­sion, includ­ing under­stand­ing ques­tions and answer­ing them based on its memory.

GPT‑3 is also able to under­stand the con­texts bet­ter which means that it could under­stand the sit­u­a­tions sur­round­ing a cer­tain word or phrase and con­vey its depth of knowl­edge or wis­dom impar­tial­ly to any­one. The abil­i­ty is real­ly use­ful in resolv­ing many com­plex issues such as those relat­ing to ter­ror­ism, drug traf­fick­ing, crime and preda­to­ry businesses.

But one bad thing about this algo­rithm is that it can be biased or have spe­cif­ic pref­er­ences or even inter­ests. For exam­ple, when asked what he thought about GPT‑3’s answers, Den­nett him­self said, “Most of the machine answers were pret­ty good, but a few were non­sense or obvi­ous fail­ures to get any­thing about my views and argu­ments correct”.

Also Read:Google’s AI “Parti”: It relies on 20 billion inputs to create photorealistic images

If AI in the future will be designed to deliv­er on its own specifics, it can poten­tial­ly become an instru­ment for manip­u­la­tion and con­trol.

We should raise our con­cerns about AI, and we should work hard to make sure that the tech­nol­o­gy does not get abused by gov­ern­ments or busi­ness­es that have ulte­ri­or motives. How­ev­er, we do not have a choice but to keep it in mind as AI is here to stay! As long as we are intel­li­gent enough to use it prop­er­ly and safely.

GPT‑2 vs GPT‑3

GPT‑2 is an unsu­per­vised deep learn­ing trans­former-based lan­guage mod­el cre­at­ed by Ope­nAI back in Feb­ru­ary 2019 for the sin­gle pur­pose of pre­dict­ing the next word(s) in a sen­tence. The mod­el is open source, and is trained on over 1.5 bil­lion para­me­ters in order to gen­er­ate the next sequence of text for a giv­en sen­tence. GPT‑2 is 10x the para­me­ters and 10x the data of its pre­de­ces­sor GPT.

On the oth­er hand, GPT‑3’s deep learn­ing neur­al net­work is a mod­el with over 175 bil­lion machine learn­ing para­me­ters. To put things into scale, the largest trained lan­guage mod­el before GPT‑3 was Microsoft’s Tur­ing NLG mod­el, which had 10 bil­lion para­me­ters. As of ear­ly 2021, GPT‑3 is the largest neur­al net­work ever produced.

GPT‑2 was known to have poor per­for­mance when giv­en tasks in spe­cial­ized areas such as music and sto­ry­telling. GPT‑3 can now go fur­ther with tasks such as answer­ing ques­tions, writ­ing essays, text sum­ma­riza­tion, lan­guage trans­la­tion, and gen­er­at­ing com­put­er code.

GPT‑3 is the most pow­er­ful and advanced text auto­com­plete pro­gram so far. It smart­ly spots the pat­terns and pos­si­bil­i­ties in huge data sets. By using that, it per­forms very amaz­ing tasks that were impos­si­ble till now by an AI tool. Accord­ing to The Verge, “The dataset that the GPT‑3 was trained on was mammoth”.

GPT‑3 vs the future

It is clear that GPT‑3 is show­ing us a glimpse of the future of AI and one thing is cer­tain that it will only be a mat­ter of time when it will make anoth­er mile­stone and sur­pass Human cre­ativ­i­ty and imagination.

The cur­rent AI are bet­ter at read­ing than writ­ing and com­pre­hend­ing infor­ma­tion than gen­er­at­ing. GPT‑3 has cer­tain­ly pushed that lim­i­ta­tion up to an extent.

The future of arti­fi­cial intel­li­gence will be man­aged by the future ver­sions of “GPT”. The future of AI will no longer be about a human to machine inter­face. It will be entire­ly a robot talk­ing to anoth­er robot to cre­ate a human-like robot.

How GPT‑3 is showing us a glimpse of the “AI Future”?

Well, the cur­rent GPT‑3 can pro­duce texts. But inevitably, the out­put will just not be lim­it­ed to text. In the future, who knows AI will give out­put in form of speech, or maybe in form of physics. You nev­er know!

GPT‑3 has already equalled a humans philoso­pher. This is not a small achieve­ment. This is some­thing that wipes off the win­dow glass­es to show us a clear­er future with AI.

Philoso­phers are hat­ing GPT‑3, though; not gonna lie. Who would have thought that AI would be tak­ing the jobs of a Philoso­pher before any­thing else.

But jokes aside, we have to be respon­si­ble and care­ful while using GPT‑3 and oth­er forms of AI. We must not for­get that tech­nol­o­gy is pulling us to the future and it is up to us what we do with it.

So, if you think AI is the future, keep your eyes wide open to pos­si­ble dan­gers and errors in the sys­tem. But if you can com­fort­ably accept and wel­come this tech­no­log­i­cal advance­ment, then car­ry on in your life with­out wor­ry­ing. What­ev­er hap­pens, we need to embrace it and find a way to make it work for us!

It can be seen that AI is not only the future of tech­nol­o­gy but also an inte­gral part of the future.

But this is only pos­si­ble if we choose to be intel­li­gent enough to take care of its side effects. With GPT‑3 as well as oth­er tools, we as indi­vid­u­als can be able to adapt with the changes but all togeth­er has to come up with a deci­sion as a whole community.

To con­clude, GPT‑3 is not here in the present form for­ev­er. It will under­go trans­for­ma­tion, ren­o­va­tion and fur­ther devel­op­ment along with time.

Leave a Reply

Your email address will not be published.

Join our NewsletterDaily Glimple of Future

Our blog, "Daily Glimpse of Future", strives to make the future much clearer than it is today. Join our newsletter for free now.