According to a recent report in the New Scientist, in less than two years (if not already), it will be possible to talk in another language using your smart phone.

After selecting the language you will be able to talk into your phone and the other party will hear the translation in his ear-piece, when the other party replies in his own language you will hear the translation in your ear piece.

Also see here

The language topic selection at Lernu does not seem to allow, the selection of more than one language. Please assume that replies in Esperanto or English are appropriate.

Jes kaj ne. Certe la Babela fiŝo alvenas kaj anstataŭigas parton de neceso de lernado. Mi povas imagi, ke oni uzas la fiŝon unue por simplaj taskoj, aĉeti kafon/bieron, demandi pri vojo ktp. Se oni parolas pri siaj sentoj, intime, uzi poŝtelefonon aŭ ia alian aparaton sentas disturbante kaj iomete false. Por diskuti pri arto kaj scienco tiel ke oni konservas nuancojn kaj precizecon, oni ne povas uzi la fiŝon dum pluraj sekvaj jaroj.

Dependas ankaŭ de la lingvoj. Kelkaj lingvoj konvenas certe tre facile, aliaj ne por aŭtomata translado. Mi vetas, ke la unuaj lingvoj estas hindoeŭropaj. Mi vetas ankaŭ, ke ni aŭdas multajn amuzajn rakontojn, kiel la translado ne funkcias, eraras.

Following story is allegerdy true.

A long time ago the software manufacturer Deplhi translated a user manual to German. To save costs Delphi decided to use automated translation. A translation about some user interface thing contained the term "scroll bar", which was translated to "Rollenbalken". So far everything was ok.

Now this material was reused in Germany and another piece of documentation was produced. This was so good that the HQ in the USA wanted to use it also. The text was translated to English, and "Rollenbalken" became "roll beam". Eh, slightly off.

The German subsidiary wanted to have a look at the USA produced version, so it was once again translated. And now "roll beam" became "Brötchenstrahl" and everyone in Germany wondered wtf???

sudanglo:The language topic selection at Lernu does not seem to allow the selection of more than one language.

Jes, pro tio ke havi mult-lingvajn fadenojn tutevidente estas tre malbona ideo. Oni markas fadenon per iu specifa lingvo por ebligi trovon de kompreneblaj afiŝoj. Se marki fadenojn per pluraj lingvoj, tio klare ne plu eblas, ĉar estos afiŝoj ne kompreneblaj. Marki fadenon Esperanta, kaj skribi la unan afiŝon en la angla, ankaŭ helpas al neniu. Estas mirinde al mi ke post tiom ĉi da jaroj, vi ankoraŭ havas abelon en la kufo pri tiu afero.

Ĉiukaze por respondi al la demando, ne, mi ne kredas ke pliboniĝoj en AI-tradukado (ege troigata kaj miskomprenata kiel ili ŝajnas esti nuntempe) kondukos al morto de Esperanto aŭ lingvolernado pli ĝenerale. Eble tio okazus se ni havus perfekte funkciantajn kibernetik-enmetaĵojn en niaj oreloj aŭ eĉ cerboj (kio por almenaŭ mi estus nepra hororo malutopieca), sed tio ŝajnas neniel baldaŭa. Ĝis tiam estas multaj kialoj kaj motivoj per lerni lingvojn, kaj fakte en eĉ la malutopia situacio de interkonektitaj cerboj mi suspektas ke sufiĉe (tamen multe malpli) da homoj lernus lingvojn. Sed tio estas paroli pri la estonteco, kaj tio kutime estas vana okupo.

Tradukiloj estas malproksime de perfekta. Por mia propra komforto, mi uzas Google Translate. Por tiuj el vi, kiuj progresas en la esperanta lingvo, estas tre evidenta en ĉi tiu ekzemplo. Aliflanke, estas certa emocio aŭ ĝojo havi la kapablon paroli al iu en dua lingvo.

Tom:havi mult-lingvajn fadenojn tutevidente estas tre malbona ideo

Mi nur volis povi indiki ke se la leganto ne scipovas sufiĉe la angla por skribi en tiu lingvo, tiam tiu povas senĝene afiŝi en Esperanto. Mi ne (evidente) volis implici ke oni intermiksu (ekzemple) la japanan kaj la francan)

Anyway, the arguments advanced so far as to why one might safely ignore the current technological developments (or as to why they are of no consequence for Esperanto) are:

1. AI translation is imperfect
2. The use of the technology is too inconvenient.

However, AI translation is advancing at a very rapid pace and the streets are already swarming with youths with smartphones wearing earpieces.

It seems clear that one can no longer market Esperanto as a 'solution' to the language problem, not just because of the current dominance of English as a lingua franca, but also because of how ridiculous it will seem to expect people to spend six months learning Esperanto when all they need to do is download an app.

Sed mi pensas ke kvalito de traduko ne estos bona.

sudanglo:1. AI translation is imperfect
2. The use of the technology is too inconvenient.

Ad 1: Not only AI translation is imperfect, all translation is imperfect. Every translation is only an approximation, biased by the translator's interpretation of the input, his capabilities, his experience and his personal bias as to how he wants the translation to be understood. So this argument doesn't really hold.

Ad 2: I would not second that. My son is a digital-native as they are called nowadays and I have never heard him complain about technology being inconvenient. For instance he prefers to read books on an ebook reader rather than reading a paper copy, due to the very simple argument that he says that books weigh more than the ebook reader. The whole disk world series by Terry Pratchet weighs several pounds, and an ebook reader with the same content still weighs less than one pound.

One aspect one might want to consider is accessability, especially due to socio-economic factors. If I hire a translator for a two day event in another language, I can easily buy for the same money some cheap android phone with google translate. And after these two days, I can keep the phone, but I have to set free the translator... If I had to spend money (and time) on a course for learning that foreign language (or Esperanto for that matter), the android phone would still be cheaper. Sure, the machine translation wouldn't be perfect, but no human translation would be either (no matter if I hire a translator or if I study the language and try to translate myself).

1. The quality of translation is one of those things that lay on scale, it isn't an on/off thing. If the translation is too much/enough off, then the AI translation (or any for that matter) is not used. There are already applications for mobile phones that translate, both speech and written text (for computers there have already been such applications quite some time). (No, I haven't tested myself.) The quality of translation of speech is said to have some way to go to be on an acceptable level, but translation of simple texts is usable right now.

On the fly speech translation will certainly fail, if you quote Sharespeare, you speak dialectly, you're drunken or simply either language (source or target) isn't among the hundred most spoken languages on this globe.

One are where translation of texts will be a hit, is touristing in China, Japan and Korea, where you can point your phone to a road, shop etc. sign and have it translated to English. And vice versa for tourists from those countries.

So it all depends on the context (as translators so eagerly always point out).

2. Digital-nativity is over-advertised. I have two teenagers, I know their friends and... Well, let's put it this way, that they do know how to watch Youtube, listen to Spotify and play Angry birds, but saving photos from the phone to a cloud service so that they could photoshop them on computer or sharing homework from a cloud service with their teacher isn't that clear at all.

Of course the usability of applications will improve, but there will be situations were any device, except a Babel fish, will be disturbing. Try to convince someone to have sex with you while your sweet talk is translated with a phone you're holding on your hand… Good luck!